This textfile generated by mechanical OCR of the document at
http://www.icann.org/announcements/gnso-review-report-sep06.pdf
Accuracy is highly doubtful.


|        Enterprise LSE
LSE Public Policy Group
LSE Public Policy Group and Enterprise LSE
A Review of the Generic Names
Supporting Organization (GNSO)
for the Internet Corporation for Assigned
Names and Numbers (ICANN)
Main Report
September 2006

LSE Public Policy Group                                                              GNSO Review
O 2002 LSE Public Policy Group
Published: LSE Public Policy Group
Contact: LSE Public Policy Group, London School of Economics, Houghton Street,
LONDON WCZA ZAE, England.
I-SE Public Policy Group
The LSE Public Policy Group (PPG) Was founded in 1998. Since then, the Work of the
group has grown considerably and PPG now includes members from many departments
in the School and from other major universities. The Group conducts a mix of
commercial and pro bono activities, including consulting, research, occasional
conferences, and publications. It Works closely With Enterprise LSE, the consultancy
arm of the London School of Economics.
Contact: j.tink1er@1se.ac.uk
Web site: httpi//MM.1se.ac.uk/col1ections/LSEPub1icPo1icy/
The GNSO Review Was Written by:
Simon Bastow, Senior Research Fellow, LSE Public Policy Group
Patrick Dunleavy, Professor of Political Science and Public Policy and Chair, LSE
Public Policy Group, London School of Economics and Political Science (LSE)
Oliver Pearce, Researcher, LSE Public Policy Group
Jane Tinkler, Researcher, LSE Public Policy Group
The following post-graduate students from LSE provided research assistance for the
Review:
Andre Alves, Efe Cummings, Francois Gemenne, Xu Ke, Silke Lechner, Julia Love,
Christian Martin, Vasyl Myroshnychenko, Maha Yuones.
5

LSE Public Policy Group                                                           GNSO Review
Contents
Executive Summary                                                                 5
List of Recommendations                                                                                     9
Part 1: Introduction                                                             12
An overview of the Genexic Names Supponing Organization (GNSO)                 12
Terms of reference for Lhis Review                                                                        13
Methodology for the Review                                                                   14
Part 2: The quality of participation and representation of stakeholders
in the Internet community achieved by the GNSO and its
Constituencies                                                                                   17
How actively do stakeholdexs panicipate in policy development?                      IS
How effectively do Constituencies represent their membexs' interests?                21
How Well do the current Constituencies represent the divexsity of stakeholders in their
areas of interest?                                                                                           30
Part 3: The visibility, transparency and openness of the GNSO process
45
The extemal visibility of GNSO                                                                 45
Document management                                                                                        50
The visibility of GNSO within the ICANN community                                    55
Part 4: How effective the GNSO has been in undertaldng its work and
developing policy positions                                                       63
The opemtions of the Policy Development Process                                         63
The GNSO Council's Working methods                                                       72
Voting in the GNSO Council                                                                     79
Achieving a coherent vision for policy development                                         36
Part 5: The regularity of the GNSO's operations in complying with
ICANN's Bylaws and operating procedures                           39
The impact of GNSO opemiions                                                                 94
6

LSE Public Policy Group                                                           GNSO Review
Annexes
A. Supporting graphs and tables
B. Summary table of GNSO Constituencies
C. Case studies comparator organizations
D. Methodology
E. Online survey results 7 Constituency members
F.  Online survey results 7 Individuals
G. Online survey results 7 Non-members
H. Terms ofreference for the GNSO Review
I.  Bibliography
7

LSE Public Policy Group                                                           GNSO Review
Executive Summary
1. Scope oflhe Review. The Irttemet Corpomtion for Assigned Names and Numbers
(ICANN) seeks to put in place aunjque model ofgovemance for the I.ntemet's
domain name system, one that rests on `bottom-up stakeholder involvement'.
ICANN has several Supporting Organizations that form a key part of this model,
along With a statement of values underpinning how this system should operate (see
Annex Figure A3). The Supporting Organizations make possible the policy
development processes Which provide the foundztions for ICANN's legitimacy as
an open and global policy-making body for the lntemet.
2. One of these bodies, the Generic Names Supporting Organization (GNSO) plays
akey role in relation to policy development about generic domain names (such as
.com, .net, .in.fo, .biz etc). The GNSO is arelatively new body, but inthe space of a
few years it has responded to mpid changes in the operations and stakeholders of
the lntemet. It has successfully generated a set of institutions and procedures for
policy development on generic names issues, and has generated policy on awide
mnge ofissues involving complicated and otien technical issues, such as access to
personal data, integrity of domain names, and procedures for growing the gTLD
space. This study reviews the operations of the GNSO in terms of their
representativeness, tmnsparency, effectiveness and compliance With ICANN's
Bylaws.
3. Represemaifveness. There are six GNSO Constituencies that Iirms, other
organizations or individuals in the area of generic names may join as members. The
Constituencies and their councilors on the GNSO Council undertake a large amount
ofwork to do With policy development on generic domain names. The
Constituencies are necessarily diverse in their nature and activity, and genemlly
take the view that a reasonable amount of autonomy from ICANN staff structures
are an important component of their bottom-up consultation Work. The
Constituencies show a mixed pattern of participation, With relatively high levels of
involvement in two Constituencies covering Registries and Registmrs, but relatively
S

LSE Public Policy Group                                                           GNSO Review
narrow pmticipation in four others, covering business users, intellectual property,
intemet service providers and non-commercial users.
4. The current pattern of Constituencies is relatively complex and no longer seems
Well-adapted to the needs of all stakeholders in the mpidly changing Intemet
community. Although the Constituency structure does provide apotential home for
almost all types of interest, there are signs that the current structures tend to reflect
a snapshot of interests that Were present at the beginning of this decade and lack
internal flexibility to incorporate new types of stakeholders from commercial and
civil society. There is consequently much scope to grow and diversify membership
of the GNSO, and to adapt structures in a such a Way that they are flexible and agile
enough to respond to new policy development issues. There are some Worrying
signs of dominance of some constituencies by a small core people and of low
participation rates in policy development Work by Constituency members.
5.  Transparency. ICANN itself is ahighly visible international body and its
decisions and activities are much discussed in the Irttemet community. However,
the external visibility of the GNSO Council is poor, largely because ofpast
inadequacies in the ICANN Website. Potential members of ICANN With interests in
generic domain names currently have to join sub-organizations (GNSO
constituencies) mther than being able to join ICANN itself. Yet GNSO
Constituencies are even less visible intemationahy than GNSO itself. So joining a
Constituency has unacceptably high information costs for anyone Who is not
already a deep insider in ICANN. This presents considerable baniers to a
fimctionjng and diversitied bottom-up policy development process. The processes
and policy development exchanges of the GNSO Council are highly tmnsparertt,
more so than most similar organizations. There are however some signs that
Constituencies are hard to penetrate for newcomers and that baseline standards such
as disclosure of interests are not adequately enforced.
6. Emclweness. The Work of the GNSO Council focuses on formally designated
`policy development processes' (PDPs) Whose supposedly mpid timings are laid
down in ICANN's Bylaws, timings Which it has not proved practicable to adhere to.
Many PDPs take quite a long time to complete and their impacts are not easy to
9

LSE Public Policy Group                                                                GNSO Review
assess. Council members devote huge amounts of unpaid time to its delibemtions
With face-to-tbce meetings, many conference calls and much emajl business. The
GNSO Council has a `legislative' pattem of operating With frequent votes, While
task forces have become essentially only sub-committees of Council membexs. The
process ofreachjng `consensus' on mqjor policy issues is otien arduous because of
conflicting interests and Weak stmctuml incentive for Constituencies to identify
core issues early and Work deliberatively to agree Widely acceptable positions. The
cument armngements for voting introduce fhrther complexities by assigning double-
Weight votes to two Constituencies (Registties and Registrars).
7. Compliance. Apart from the unreahstic timings for policy development process, the
GNSO's operations comply With the ICANN Bylaws. There is however relatively little
sign that policies developed by the GNSO since its establishment have been subject to
comprehensive impact assessment.
S. Princxplesfor making change:. Any changes made to the GNSO's opemtions need to
follow through on four key pxinciples:
-    The GNSO's opemtions need to become more visible and tmnsparent to
a Wider mnge of stakeholders than at present.
-    Any reforms made need to enhance the representativeness of the GNSO
Council and its Constituencies.
-   The GNSO's structures need to be more flexible and adaptable, able to
respond more effectively to the needs of new and old stakeholdexs in a
rapidly changing Intemet environment.
-   Changes in the GNSO Council's operations are needed to enhance its
ability to reach genuinely consensus positions, ergoying Wide support in
the Intemet community.
9. Specmc suggestion: for reform.  We foxmulate a set of 24 evidence-based and
pmcticable recommendations to help GNSO to improve Where there are cutrently
problems. These suggestions can be accepted or not i.ndividuz].ly, but they hang
together as a coherent body of reforms. Some main points include:
-   cutting down the number of Constituencies from six to three, covexing
registration interests, business users and civil society;
10

LSE Public Policy Group                                                                GNSO Review
-    creating a direct (pximary) membexship in ICANN for tixms, other
organizations and individuals. Newly joined members interested in
genexic names issues Would then be directed to also join one of the new,
simpler and eager to understand Constituencies that We outline below.
The Constituencies Would receive more ICANN support to sustain their
activities and outreach Work. While being run by and accountable to their
members as now;
-    creating radically improved ICANN and GNSO Websites that can
effectively represent the GNSO to the Intemet community as a Whole;
-    abolishing the current Weighted voting for registmtion interests but
giving both them and business users (broadly construed) an effective
veto over non-consensus change;
-   raising the threshold for consensus policy from 66 to 75 per cent
agreement;
-   radically reducing the use of telephone conferencing and shifting to
more face to tbce GNSO Council meetings, for Which all participants
Would receive reasonable travel and accommodation expenses;
-   making more use of intensive task forces to bxing in extemal expertise,
to broaden the involvement of interests from the Intemet community and
to speed up policy development;
-   using staff expertise more fijlly and constructively to speed up policy
development and to help focus GNSO Council's attention on making
key issues and decisions;
-    creating texm limits for GNSO councilors (of either three or four years)
and putting in place stronger protections against the non-disclosure of
interests.
1 1

LSE Public Policy Group                                                           GNSO Review
List of recommendations
(In this list the paragraph number given in black rejer to the specmc point in the main
text where theftll recommendation is spelt out and explained There is generally some
analysis ofthe rteedfor change gwen in the main text paragraph: immediately before
each recommendation).
Recommendation 1
A centralized register ofull GNSO stakeholders should be established which is up-to-
date and publicly accessible. ]t should include the member: of Constituencies and
other: involved in the GNSO tmkforces.  (Paragmph 2.5)
Recommendation 2
GNSO Constituencies should be required to show how many member: have purtictpated
in developing the policy positions they adopt. (Pamgmph 2.14)
Recommendation 3
There needs to be greater coherence and standardization across Constituency
operation:.  For this to work epectfvely, more JCANN stgf support would be rteededfor
constituencies. (Paragraph 2.22)
Recommendation 4
A GNSO Constituency support omcer should be appointed to he? Constituencies
develop their operation websites and outreach actfvity. (Pamgmph 2.23)
Recommendation 5
Constituencies shouldfoctu on growing balanced representation and active
partietpation broadly proportional to wider global distributiomfor relevant indicator:.
(Pamgmph 2.39)
Recommendation 6
The ba.vi.vforpartic?ati<m in GNSO activities need: to be revised from Constituency-
bmed membershw to one deriving from direct JCANN stakeholderparticwation.
(Pamgmph 2.44)
Recommendation 7
The GNSO should improve the design and organization ofthe current website, develop
a website stmtegyfor continual improvement and growth over the next three year:. and
review usage statistic: on a regular bent: to check that tregfe to the website it growing
over time and understand more fully what external audience: are interested in.
(Pamgmph 3.10)
Recommendation 8
Document management within the GNSO need: to be improved and the presentation of
policy development work made much more accessible. (Paragraph 3.14)
12

LSE Public Policy Group                                                                GNSO Review
Recommendation 9
The GNSO should develop and publish annually a Policy Development Plunfor the next
two year:. to act both as a strategy document for current and upcomirtg policy work
and as a communication: and marketing tool for general consumption outside ofthe
JCANN community. It should dovetail with ICANN iv budget and strategy document:.
(Pamgmph 3 . 1 6 )
Recommendation 10
The GNSO and JCANN should work proactfvely to provide information-based
incentive: for stake}/[older organizations to monitor urtdpurticwate in GNSO issue:.
(Pamgmph 3 . 1 9 )
Recommendation 1]
The position ofthe GNSO Council Chair need: to become much more visible within
JCANN and to Carry more institutional weight. (Paragraph 3.26)
Recommendation 12
The policies on GNSO Cozmcilor: declaring interest: should be strengthened. Provision
for avote of (no corjtdencey leading to resignation should be introduced for mm-
compliance.
(Pamgmph 3 .23)
Recommendation 13
Fixed term limit: should be introduced for GNSO Cozmcilor: either of two two-year
term: (as applied in some Constituencies already) orperhap: ofa single three-year
term. (Paragraph 3.30)
Recommendation 14
The GNSO Council and related policy stqf should work more closely together to grow
the use ofproject-management methodologies in policy development work particularly
focusing on how targeted issue analysis can drfve data collectionjtom stakeholders
(rather than vice versa). (Pamgmph 4.14)
Recommendation 15
The GNSO Council should rely more onface-to-face meeting: supplemented by online
collabomtfve method: of working The Chair should seek to reduce the use of whole-
Council teleconjhencing  (Paragraph 4.19)
Recommendation 16
The GNSO Cozmcilor: should have access to afzmdfor reasonable tmvel and
accommodation expense: to attend designated Council meetingg instead oflmving to
meet such costsjtom their own resource: as at present. (Pamgmph4.21)
Recommendation 17
The GNSO Council should make more use ofTa.vk Force:. Task Force pmticwunt:
should be more dfverse and should be dmwnjtom a wider range of people in the
Jmemet community, and national and international policy-making communities.
(Pamgmph 4 .26)
13

LSE Public Policy Group                                                           GNSO Review
Recommendation 18
An ]CA]VNA::ociute stakeholder Category ofpurticwatiort should be Created :0 as to
create a pool of readily available external expertise, which can be drawn upon to
populate Task Force: where relevant. (Paragmph 4.27)
Recommendation 19
The current GNSO Constituency structure should be radically simplmed so as to be
more capable of responding to rapid change: in the Jntemet. The Constituency
structure should be clean comprehensive (covering allpotential stakeholders) and
flexible, allowing the GNSO to respond easily to the rapid change: in the make-up of
lmemet stakeholders. We suggest a set ofthree larger Constituencies to represent
respectfvely Registration interests, Bminesses and Cfvil Society. (Paragraph 4.35)
Recommendation 20
A reorganization ofGNSO Constituencies would also allow the Council to be made
somewhat smaller (we suggest 16 member.? and hence easier to manage. (Paragraph
4.36)
Recommendation 2]
The dejtnitiort of achieving a consensus should be mixed to 75 per cent. Weighted
voting should be abolished. Both measure: could he? to create more irtcerttfvesfor
dwrmnt Constituencies to engage Constructively with each ot}/ten rather than simply
reiterating a (bloc 7 position in hope: ofpickirtg up enough uncommitted vote: :0 as to
win. (Pamgraph 4.33)
Recommendation 22
The way in which the GNSO Council vote: to elect two Director: to the JCAMVBOWJ
should be churtgedto use the Supplementary Vote system. (Paragraph 4.40)
Recommendation 23
The Mount ofdetailedprescrwttve provision in the JCANN Bylaw: relating to the
operation: ofthe GNSO should be reduced JCANN Bylaw: should outline broad
prirtcwle: and objectivesfor the GNSO but the detailed operational provision
mtcluding the section on the PDP) should be tmmjlrred to the GNSORule: of
Procedure. This would allow the GNSO to agree amendment: and to introduce new
innovation: in its working methods and timeline: in a more realistic wtdjlexible way,
while operating within JCANNX: guiding princwles. (Paragraph 5.7)
Recommendation 24
Both JCANN and the GNSO Council should periodically (my once everyffve year.`?
compile or commission uformul (quwttitatfve and qualitative) assessment ofthe
influence ofthe GNSOX: work on developing policy for generic name:. This should
include an analyst: ofhow the GNSOX: influence with national government.`,
international bodies and the commercial sector might be extended (Paragraph 5.12)
14

LSE Public Policy Group                                                           GNSO Review
Part 1: Introduction
An overview ofthe Genetic Names Supporting Organization (GNSO)
1.1 The Intemet Corporation for Assigned Names and Numbexs (ICANN) Was
established in September 1993 as a non-profit corporation responsible for the
assignment of unique identitiers making up the Intemet domain name and numbexing
system, and the technical coordination required to ensure the system's stability and
interoperability. A mqjor pan of ICANN's mission is to develop global policy relating
to its technical and management functions in the genexic and countxy code top-level
domain (TLD) systems. Its cument structure consists of three Supporting Organizations
responsible for the policy development role. (See Annex A. Figures A1 and A2 for
more information on ICANN 's internal organization.)
1.2 In ICANN terminology a Supporting Organization (SO) is a consultative and
policy-development body, Whose iimction is to allow multiple stakeholdexs in the
global Intemet community to conttibute to policy-making on matters that fbll Within
ICANN's remit. The views of SOs go to the ICANN Board, and Where the SO can
achieve a consensus their view has special force in guiding and shaping Board policy. It
has been an impoxtant part of ICANN 's distinctive character as the body guiding
Intemet development in respect of domain names and numbers that SOs allow for
bottom-up involvement by diverse stakeholders. SOs also play a key role in fostexing
the development of consensus policies, i.e. those ertjoying a broad and substantial level
of agreement amongst different interests and communities involved in the Intemet (even
if not always univexsal agreement).
1.3 The GNSO Was set up in December 2002. It is responsible for developing and
recommending to the ICANN Board substantive policies relating to genexic top-level
domains. The GNSO consists of six Constituencies designed to represent the interests of
different groups of stakeholdexs in genexic names: gTLD Registties, Registrars,
Business and Commercial Users, Intellectual Propexty, Intemet Service and
Connectivity Providers, and Non-Commercial Usexs. Each Constituency recmits
membexs and armnges to consult them, usually establishing a chair and executive
structures to help process GNSO business and to collate the views of Constituency
15

LSE Public Policy Group                                                           GNSO Review
members. The Constituencies are summarized in Figure 1 below and fimher
information is contained in Annex B at the end of the report.
Figure 1: Summary of the six GNSO Constituencies
Cunstituemy name                            Whu the Cunstituemy mainly represenh`
gTLD Registries                           Registries are tirms that operate top level
domains and provide connections to the
I.ntemet's root sewers. This Constituency
also includes `sponsors' (organizations
representing user communities to Which
ICANN has delegated rights to manage top
level domains .
Registrars                                   Registrars are Iirms marketing the
registmtion of domain names to Iinal
customers, businesses and other users
Business and Commercial Users         Corporations, small and medium enterprises
and trade and professional associations
re|resenti ; such co |orations and SMEs
Intellectual Property                        Tmde associations in the US, Europe and
intemationa].ly Who monitor intellectual
property (IP) rights issues and
infri |ements, and la   ers inthe ]P area
Intemet Service and Connectivity      Firms marketing Intemet connectivity, emajl
Providers                                     services and alien Web site domains to final
customers
Non-commercial Users                     Noncommercial organizations With
registered domain names in gTLDs or
ccTLDs, such as the universities, charities
and NGOs
Each Constituency elects three members to the GNSO Council. In addition, three other
members of the Council are appointed by another part of the ICANN organization
called the Nominating Committee (Whose role is to bring into ICANN talented people
With a disinterested stance). The GNSO Council thus has 21 members and its operations
are the core of the Supporting Organization. The Council meets three times a year, tbce
to fbce, at the ICANN meetings.
Terms of reference for this Review
1.4 Since December 2002, ICANN's Bylaws have stated that periodic reviews should
take place of its SOs to examine how their procedures are Working. Following
competitive tender, in February 2005 the ICANN Board commissioned the LSE Public
16

LSE Public Policy Group                                                                GNSO Review
Policy Group to conduct an independent review of the GNSO. The ICANN Bylaws
(Article IV, Section 4) stipulate that two high-level questions should form the bags for
the review of its supporting organizations as follows:
-    Whether the organization has a continuing putpose in the ICANN structure;
and
-   Whether any change in the structure or opemtions is desimble to improve its
effectiveness.
The ICANN Board additionally asked the LSE Public Policy Group to look at the
following four issues:
-    Representativeness;
-    Transparency;
-    Effectiveness; and
-    Compliance.
We have grouped the exact specifications in the Terms of Reference by each of these
four broad issues. Annex H provides a 6].1.1 listing of the Texms of Reference, arranged
by these four grouped areas.
Methodology for the review
1.5 To conduct this enqlmy We used a mnge of different methods, designed to generate
awealth of objective evidence, and to yield different kinds of data that `ttiangulate'
With each other. Txiangulation is a technique Where the inevitable limitations on the
useiijlness of any one method; can be countered by dmwing on the strengths of other
methods so as to yield more reliable judgments and estimations. Our approach is set out
in detail in Annex D of this report, but bxiefly our main methods Were:
-   From the extensive document archives on the ICANN Website, and With the
generous assistance of ICANN staff We assembled a thorough documentation
of GNSO's activity since its formation in 2002.
-   We collated data on the conduct of GNSO's policy development process,
looked at the issues and debates covered and examined the viewpoints
expressed in the GNSO Council by Constituency representatives. We also
collated data on GNSO Council votes.
17

LSE Public Policy Group                                                           GNSO Review
-   We examined all the documentation available on GNSO Constituencies,
looking at membership details, paxticipation in policy consultation, and the
involvement of pexsormel.
-   We established a Website for the GNSO Review and Wrote to all Constituency
membexs that We could identify asking them to give us their views in response
to an on]j.ne suxvey posted there, and to send us views by emajl or contact us
to give their views in pexson. In the review pexiod 107 people filled in the
survey and a considerable number of people contacted us additionally.
-    We conducted lengthy and detailed face-to-tbce or phone intexviews With over
100 people Who Were expexienced and knowledgeable about GNSO's role in
ICANN, including all but one current GNSO Council members and relevant
membexs ofthe ICANN Board.
-    We visited the ICANN meeting in Wellington and obsexved arange of
different meetings of Constituencies, meetings of GNSO Council and public
presentations by GNSO Council and the ICANN Board. We also obsexved a
phone conference of GNSO Council.
-   We systematically scanned the Web for comments or discussions on the Work
of GNSO and ICANN.
1.6 The opemtions of GNSO are of centml impoxtance for those of ICANN as awhole,
but it is important to note that the scope of our enquiry Was ]j.mited to GNSO alone.
However, some of our Endings and recommendations below have a certain degree of
broader relevance. They may mise some issues With implications for the other parts of
ICANN With Whom GNSO ]j.n.ks and interacts.
1.7 In our onhne sutvey, We asked respondents to score a mnge of different challenges
thcing the GNSO in the future in terms ofimpoxtance. We used a Liken scale from 1 to
7, Where 1 | Not at all important to 7 | Very important. Figure 2 below gives a
summaly of the number of times different challenges mnked highest in people's
selections and the number of time different challenges mnked lowest. The net score
gives some indication of overall importance. The most important challenges for the
GNSO involve improving the quahty ofgTLD policy making, and improving
tmnsparency and openness.
IS

LSE Public Policy Group                                                           GNSO Review
Figure 2: Survey respondents' views on the main challenges facing the GNSO in
the future
Ranked   Ranked     Net score
highest    lowest      (highest-
lowest
lm| o ~ |the |uzlit of|TLD |oljc maki ;         35      17        IS
Improving Lmnsparency and openness in g'I'LD             40          22              IS
|oljc  develo|   ent
Representing more effectively the views of the             36          24             12
Imemet users Worldwide
Broadening the mnge of organizations panicipating       23          23               5
in |TLD |olic  develo a   ent
Some other challen|e                                            7          2              5
Encouraging more intensive panicipaiion by mqj or        22          25              _ 3
o |anizationsin |TLD |olic develo|ment
Raising the projile of the GNSO as apolicy                  24          34             -10
develo a  ent bad
was  Raw; m blank fm was mm Cnmmnnly Yanked mggm than lnwest as a pnnnty  Raw; m ma gum Was mm Been gem as
alnwpnnnty
same Lse ppa Drama gm| Dfcnnsumem members anamamauaxs
1.8 The Review is structured into four main substantive Pans looking at:
-    The quahty of the panicipaiion and representation of stakeholders in the
Intemet community achieved by the GNSO and its six Constituencies (Pan 2);
-   The visibility, Lmnsparency and openness of the GNSO process (Pan 3);
-    How effective the GNSO has been in undenaking its Work and developing
policy positions (Pan 4); and
-    The regulaxity of GNSO's operations in compliance With ICANN's Bylaws
(effectively its intemal constitution) and the implementation of GNSO policy
positions (Pan 5).
19

LSE Public Policy Group                                                           GNSO Review
Part 2:
The quality of the representation of stakeholders in the
Internet community achieved by the GNSO and its six
Constituencies
2.1 This Part of the report examines the quzhty of participation and representation
achieved by the GNSO and its six Constituencies. It focuses on the following questions:
-   How actively do stakeholders participate in policy development?
-   How effectively do Constituencies represent their members' interests?
-   How Well do the current Constituencies represent the diversity of stakeholders
they claim to represent?
-   How can the GNSO further enhance participation and representation?
2.2 Achieving a reliable comparative picture of representation and participation across
the Constituencies has not been particularly easy. Our interviews have generated a Wide
mnge of views about practices and activities in other Constituencies. We have ottert
found that strong perceptions prevail across Constituency members about the quahty of
participation and representation in other Constituencies. At times, perceptions have
seemed quite predictable in that certain groupings of Constituencies tend to hold quite
entrenched views about other groupings. Our unobtrusive analysis of Constituency data
has also generated a great diversity of information about the nature and extent of
participation and representation across Constituencies. As With all our tirtdings in this
report, We have sought to triangulate different types of evidence as thoroughly as
possible in order to control for misleading perceptions.
2.3 The existence of strong perceptions across Constituencies is perhaps an inevitable
timctiort of the strong and diverse interests at stake in the GNSO (particularly business
interests). There is a danger that perceptions put forward by competing interests Work
With equal and opposite forces, a `superstructure' consisting of all possible perceptions.
From the point of view of the researcher attempting to sort perception from reality, the
trick is to break down this superstructure and clear a space on Which to build analysis.
More importantly, from the point of view of the GNSO and its policy development
goals, this kind of self-perpetuating superstructure can have a corrosive impact on
20

LSE Public Policy Group                                                                GNSO Review
policy consensus. For there is little incentive for one Constituency to trust the behavior
of opposing Constituencies to a sudicient extent that prevailing perceptions may be put
adds. Constituencies may therefore be inclined to take up defhult positions in `equal
and opposite mode' based on prevailing perceptions, and the potential for meaningfhl
consensus will be greatly reduced. (Annex Figures A9 and AIO give abroad picture of
some of the prevailing perceptions that suxfaced, Which We do not explore in more
detail here.)
How actively do stakeholders participate in policy development?
2.4 Although the ICANN model itself is not predicated on the idea of membership, the
GNSO Constituency model does operate through amembexship system involving
relatively autonomous mechanjsms for bottom-up paxticipation. Constituencies have
considerable flexibility in how they support, and are accountable to, their members. One
minimal requirement for this system however should be that cument and potential
membexs across all Constituencies should be able to see Which other organizations are
cumently members, and should be able to get in touch With other member organizations
via designated individual representatives Who are easy to identify. We Were not clear
that this requirement is currently met. A great deal oftime has been spent duting this
research seeking to identify and follow up on organizations listed as members of the
GNSO Constituencies. The GNSO Secretaxiat kindly requested that the six
Constituencies complete a questionnaire containing core information about their
organizational structures, processes, and cument membexs. Three Constituencies
responded to this questionnaire, the GTLD Registry Constituency, the BC, and the NCU
Constituency. However it took our research team considerable time and effoxt to
compile a 6].1.1 list of membexs for all six Constituencies, and to identify individual
paxticipants and their ]j.n.ks to paxticular member organizations. Some Constituency
Websites do not even keep an up-to-date list of membexs. The current difticulty of
Ending full and accumte information on Which organizations are currently membexs of
Constituencies makes it hard for prospective membexs to determine how representative
constituencies are and how Woxthwhjle it Would be for them to paxticipate. This has
some has obvious adverse implications for transparency (see Part 3 below).
21

LSE Public Policy Group                                                                GNSO Review
Recommendation 1
A centralized register ofull GNSO stukeholders should be established which is up-to-
daie and publicly accessible. Jr should include the member: ofCon.vIiIuencie: and
other: involved in the GNSO Imkfome:.
2.5 We recommend setting up a centrally held onhne register of GNSO Constituency
stakeholders, With genexic information about member organizations showing What type
of Him or body they are, its contact details, and the name and emajl of a designated
representative. Where a stakeholder is itself a representative organization (such as an
interest group or tmde association) acting on behalf of others, some basic infoxmation
should be given about the number of organizations represented. The stakeholder register
should be publicly available on the GNSO or ICANN Website.
2.6 Tuming to the issue of how current members of Constituencies participate in their
affaits, We again found Wide dispaxities across Constituencies in the information
available on participation. Having established some kind of comprehensive spreadsheet
of Constituency members and possible designated contacts, our evidence suggests that
there are gaps of varying sizes between the organizations listed as members, and the
actual participation by these organizations in Constituency policy development. Our
researches trawled Constituency Websites and the GNSO Website for as many
indicators of participation as possible. We followed this up With a formal request to the
Constituencies for any data on participation not publicly available. Only one
Constituency, the gTLD Registxy Constituency responded to our request With detailed
information. We also received awxitten response from the BC.
2.7 There is no single, perfect Way to gauge participation in constituencies given the
diversity of the industry communities represented. One Way of gauging the level of
activity at Constituency level is to analyze the extent to Which mailing lists are used.
Some Constituencies such as the Registmr and the NCUC have open access mailing
lists, Which give a good indication of the extent of participation in the Constituency and
the mnge of organizations involved. Figure 3 shows a basic analysis of mailing lists of
four Constituencies out of six. It is useful to compare the number of contxibutoxs With
total cument members, and the number of postings by half-year pexiod. The number of
conttibutors for the supply-side Constituency lists, Registties and Registmrs, are
broadly equivalent to the number of cument members. Registry postings have increased
22

LSE Public Policy Group                                                                GNSO Review
by around 50 per cent in the pexiod 2003 to 2006, While Registmr postings to their
public list have decreased by roughly the same propoxtion (a new pxivate mailing 1ist for
Registraxs may explain this decrease). The NCU Constituency shows slightly less
activity on its public 1ist compared to the
Figure 3: Analysis of activity on Constituency public mailing liss
Current     Total           Number of          Number of            Core ratio
members   participants   conttibutoxs in      postings in half       (25 : 75)
half  ear |exiod       ear |exiod
Re|~s        14      No dam   U|, 14 to 13     U|, 700 to 1100   No dam
Registrar        56            101        Steady, 40 to 50   Down, 1000 to       25 : 73
400
NCU            44            51       Down, 23 to 17    Down, 330 to 230   25 : 71
Isp                  42       -  Stead  ,4 to s        zo to so                 _
BC               39         No dam    No dam             No dam               No dam
|     33        No dam    No dam             No dam              No dam
NOTE 1 The figures m ms table summarize more detaned data presented m Annex A, Ftgms At 1 to Am, ofmts
report The number ofpostmgs and individual contnbutors are gwen for half year pmods from January zoo; so June
zoos Tm! participants show the total number of mdmduals contnbuttng to the nst at least once dunng thts ttme
penod Number at cantrmutnrs in halt year Pmna mdtcates how many mdmduals are parttctpatmg tn the nst
Number at posting; in hat: year pmna `names how much the nst `S bemg used Cure min gwes an esttmate of
the concenntton ofusage by showing the percentage omni postmgs sent by the top zs per cent contnbtws The
we zs 75 mdtcates that the top zs per cent of contnbutors account for 75 per cent oftotal posnngs
NOTE z For the NCUC, zoo; Was a period m Wmch they completely remed then charter and msnmted new
memb ashxp mtena, leadmg to a mgher than normal level of acttvxty
NOTE 3 The Be pomwd out so us that although We compare auctuattons m parttcnpatton rates m thts table, absolute
sgm-es are not strictly comparable for the purposes of drawing conctustons
same LSE pm and as Dfcnnsummc  |ubhcmanm| has
registmtion Constituencies, and had just fewer than 20 contxibutoxs to the list in the tixst
half of 2006 (equivalent to just under half of total membexship). Mailing list data Was
not available for the Business and ]P Constituencies. The actual participation levels tend
to suggest a `rule of thumb' ratio of 25:75, Where one quarter of the members are
generally responsible for at least three quzrtexs of the paxticipatory activity.
2.3 We looked for alternative Ways to evaluate the extent of participation at
Constituency level. Three out of six Constituencies provided data on the number of
paxticipants attending Constituency meetings since early 2003. The BC and the IP
Constituency publish minutes of meetings on their Websites. We received data from the
Registry Constituency on meeting attendance. Figure 4 shows the Endings. Again the
Registry Constituency shows solid attendance in relation to total members over 1 15
meetings for Which data Was available. The BC and1P Constituencies also show steady
attendance even if the regulaxity of the meetings held appears to be much 1ess than the
23

LSE Public Policy Group                                                                GNSO Review
Registry Constituency. We Were not able to tind any data on meeting attendance for the
other three Constituencies, either on their Websites or in response to formal requests.
Figure 4: Analysis of average levels of attendance at Constituency meetings
Current     Total             Avemge           Average           Meetings for
members   participants      paticipants at     paxticipants at     Which data
face-to-face          teleconference      Were
meetin|s              meet  |s               available
Re|~s         13      Mm/mlnble | 10 to 17     Stead 10 to 12      115
|      33               133          Steady around     From 40 to to              13
20
BC           39          51      Stead 10to 12  Stead 10 to 12       21
Re|gtrar           56              No data         NO data                  No data                         Nil
NCU                44              No data        NO data                 No data                        Nj]
Isp                   42              No data        NO data                 No data                        Nj]
Note: The jigures in this table summarize more detailed data presented in Annex A, Figures A15 to A18, of this
report. Participation is calculated for all meetings for which minutes were available, and jigures show average
number of parti cipants per meeting for all years available. Total psrticip ants show the total number of
individuals attending any meeting at least once since the start ofZ003.
same LSE ppa and as Df |ubhd  avattamemmutes Dfcnnsummc  mean|s
How effectively do Constituencies represent their membexs) interests?
2.9 Our on]j.ne survey sought to capture the views of Constituency member
organizations on how Well represented they felt by their respective Constituency. In all,
51 current membexs of Constituencies (around one titth of the total) took the trouble to
complete a sutvey. Although this response rate makes it difficult to dmw rock-solid
conclusions for individual Constituencies, it is possible to draw some limited
conclusions, especially if there are any signs of dissatisfaction With the Constituency
process.
2.10 Amongst respondents Figure 5 shows that there seems to be general satisfhction
With the Way in Which Constituencies represent their members' interests. All ten
Registry respondents rated their Constituency at live or above, reflecting a general
feeling that Was buttressed in interviews that the Constituency is largely representative.
In other Constituencies, although there is general satistbction, there are also signs that
some members feel there is scope for improvement in the Way that their Constituency
represents the Wider army of organizations oftheir type. Our interviews With
Constituency membexs coniitmed these signs of dissatisfaction around the edges.
24

LSE Public Policy Group                                                           GNSO Review
Smaller registmrs We spoke to suggested that the Registmr Constituency tends to be
more representative of larger registmrs and could do more to encoumge membership
across the smaller organizations. Some members of the BC also suggested that the
influence of one or two individuals tended to detmct from the overall quahty of
representation in the Constituency.
Figure 5: Survey respondents' views on how effectively their Constituencies take
account of their views
Registry   Registmr   NCU      IP       BC       ISP
Scwsgwenpy
How effectivel    our Constituenc   takes account of the views of  our o |anization
I to 3 |       2           2 |     I
4                                       _       I |
5 to 7                                       to            11           7         4 |
How effectivel    our Constituenc   takes account of all o |anizations of  our t An|
I to 3 |       4           I |     2
4 |       I           2 |
5 to 7                                       10      18**    4         5
NOTE: Mm details on the online suwey questions can be fountain Annex E. Respondents wee askedto
give a score for each ofthese questions on a Liken scale from 1 to 7, where 1 I Not at an effectively and
7 I Very effectively. we receivedinsuaicient responses Rom the Isp Constituency.
sum tsnppamumem| Dfcnnsumenc members
2.11 Looking in more detail at satisfaction levels amongst Constituency member
organizations, We asked a series of questions in our suxvey about the Way in Which
respective Constituencies develop policy positions. There are some interesting
vaxiations in Figure 6, paxticularly in the relative scores Within Constituencies. The
highest scaling aspect across all Constituencies is for the production of authoritative
Wxitten statements of Constituency positions. Apalt from the Registxy Constituency
Which scored distinctly highly across all three questions here, respondents tended to
score their Constituencies compamtively lower on making data available on the degree
of consensus across Constituency members. The Registmr Constituency and the BC
also scored comparatively lower on the use of consistent procedures to canvas views
across their membexs.
25

LSE Public Policy Group                                                                GNSO Review
Figure 6: Survey respondents' views on how effectively their Constituencies
develop and evidence policy positions
Use consistent         Produce authoritative     Make data available
procedures for         Wxitten statements of        on the degree of
establishing consensus   Constituency positions      consensus across
across Constituency                                         Constituency members
membexs
Re ; ~ s                                           6 .4                                                   6 .3                                                  6.3
NCU                             5.4                                  5.3                                 4.3
|                 5.5                                  5 .5                                 3.5
Re|~strar            4.6                  5.3                  3.1
B C                                                4 .6                                                   5 .3                                                  3. 1
ISP                                                    Jmu|xciem dam available
Total                             5.3                                  5.5
ave . |e
NOTE: More details on the online survey questions can be foundin Annex A. Respondents were asked to
give a score for each ofthese questions on a Likert scale from 1 to 7, where 1 I Not at all effectively and
7 I Very effectively. We received insufiicient responses from the ISP Constituency.
same LSE ppa mm Sway at cnnsumemy members
2.12 There Was considemble vaxiation in number and availability of Constituency
statements and other foxmahzed policy positions. In order to assess mnge and
availability across Constituencies, We compiled a list of IS policy issues that
Constituencies might reasonably have commented on in recent yeals (see Figure 7).
These included all Constituency statements produced as pan of formal policy
development processes (PDPs), as Well other relevant issues Which Were not subject to a
PDP such as Constituency responses to the ICANN Stmtegic Plan (Annex A. Figure
A22 lists these IS policy issues in full). The Registxy Constituency pointed out that
some of its Constituency statements had not been included in our list of IS possibles.
Figure 7 also shows the extent to Which statements gave badc information on the
number of Constituency membexs participating in the development of the statement. We
have contmsted these data With another proxy for the intensity of representation and
paxticipation, the propoxtion of postings by the GNSO Counciloxs to the GNSO public
mailing list. Statements published by the Registry and Registrar Constituencies Were the
only ones that showed any sign of containing data on the number or range of
organizations participating in the development of the Constituency position. None of
the Constituencies representing Intemet users displayed any data on the number of
organizations actively participating in the development ofpoljcy positions. These
Constituencies are vexy active in the Work of the GNSO Council. For example, the BC
26

LSE Public Policy Group                                                                GNSO Review
made available a very high proportion of Constituency statements and has easily been
the most vocal Constituency on the GNSO mailing list in recent years.
Figure 7: Analysis of published Constituency statements and levels of contribution
from Constituencies on the GNSO public mailing list
Publication of                    Number of
Constituency statements          Constituency
(out of 13)            statements containing
data on Constituency
member |axtici| . ~  ;
Based on.-                                 Armexxt Fig  re A22
Re ; ~ s                                                 1 1                                                  3
Re|ttrar |                  1
Be            IS |
|                    13                   -3
NCU           12 |
ISP                    -:-3
SOME Lse pm analysts Of cnnsummcy statements pubhshea an cnnsmuemy Webms BY
pubhshedtntssuerepnns Dubs attso
NOTE! Fm am hst Dubs la pnhgytsms sewed by the tsmeam pteasmfmn Annex A
mgm Azz xt shnutdbenmaumthe pnhcytssues selected are my asubyset Dfau Msg Mused
by cnnmmennes. and Was Chasm mdepenamny by the LSE team
Now z The Regmw cnnsummcy Cmsmma that an stamens Med by the cnnmmmcy Was
Med as msuu Dfamqnntl Mewhsnmmnntnqe dmmmt Gamma mated damn
Now 3 xn we Manes. the so med mmentnwnhetrmws Was gun Cnnststmtmnnhm
nn|max mm|
2.13 The GNSO has made progress in recent policy development Work to establish a
more systematic structure to Constituency statements in texms of amore formulaic
question and answer format. We discuss in Pan 4 how the initial report in the policy
development process might be structured around akey issues analysis or issue-based
questions. We also discuss the possibility of integmting some quantitative questions
into the Constituency statement to distinguish between different pxioxities and interests
in Constituency positions. However, cutrently only a small mjnoxity of policy
statements developed by Constituencies have any kind of data indicating the level of
paxticipation or suppoxt for a particular Constituency position. One or two recent
statements from the Registry Constituency have contained data on the number of
membexs participating in the process. We have heard a mnge of arguments for and
against the introduction of some basic underlying data of this kind. Senior personnel,
including Board membexs, have suggested that more in-depth information on levels of
paxticipation and agreement from the Constituency members Would facilitate Weighing
up and intexpreting any final policy recommendations from the GNSO Council. Other
27

LSE Public Policy Group                                                           GNSO Review
interviewees argue that Constituencies are alien not required to achieve consensus
across their membexs (and at times cannot). Some have said that members cannot be
forced to participate, and so silence from member organizations on apaxticular issue
can legitimately be taken as agreement on a consultation document. There have been
attempts in some recent policy development processes to collect underlying data from
Constituencies.
Recommendation 2
GNSO Constituencies should be required to show how many member: have parlicxpaied
in developing the policy positions they adopt.
2.14 We recommend that all Constituencies should be required to poll members on
Constituency statements produced as pan of apolicy development process (as some
Constituencies already do), and Constituencies should provide a summaly ofpoll
results When submitting statements to the policy development process. Constituency
statements feeding into the policy development process should incorporate evidence of
how many members have aclfvely agreed to Constituency positions. Silence from
member organizations should not be interpreted as agreement (contmry to What some
Constituency membexs suggested to us in interviews). Flmhermore, We suggest that
Constituency membexs have a means to indicate their level of agreement on
Constituency-Wide positions, on a simple category scale such as: Strongly Agree,
Agree, Neither agree nor disagree, Disagree, Strongly disagree. This data may feed into
repoxts in an aggregated and anonymized format as an indication of the levels of
consensus views at Constituency level.
2.15 Any discussion of how effectively Constituencies are able to represent their
membexshjp should take into account the resources available to the Constituencies to
thud their operations. The GNSO Constituencies receive revenues oftheir own through
annual membexship fees paid by their membexs to cover their administmtive and
program costs. These fees vaxy considerably across the Constituencies, from less than
USDIOO to nearly USDIOOO, With a (guess)timated total revenue of around USD
35,000 across four constituencies (excluding the gTLD Registry and ISP
Constituencies, see Annex B).
23

LSE Public Policy Group                                                                GNSO Review
2.16 GNSO Councilors are not currently paid for their Work on the Council and in their
Constituency. Some Councilors may have business interests relevant to Council
discussiorrs and can sometimes be compensated for time invested in Council business
Figure S: GNSO proportion of ICANN expenditure and revenues
USD thousands                                              Fiscal year
2003-04      2004-05      2005-06    2006-07
ICANN annual expenditure
3,273         15,334       22,938       30,977
ICANN revenues
3,618     16,032    23,566   34,179
Revenue jrom gTLDRegi:xrie: and
Regislmrs as apercemage oflolul                 87             9]                             97
ICANN revenue
Eslimaied expenditure on the GNSO as
apercemage OIIOMIJCAMV                          2               4                              7
ex ' endixure
No-rn rnrmals Drhnwesumatea Expmawe rmhe awso Was Calculated please seemenntes tm rrgummu rn AMEXA
xcmm annual Expmame anamvenue sgums fm znmym, znnws, and znnsyna are appmvea budget sgms rmm xcmm
rmanmax Ra|m warm zuuaym are |m|used bur|st s|ums
and cover Lmvel costs through their employers. Members of the Council appointed by
the Nominating Committee receive travel costs for attending ICANN and other GNSO-
related meetings.
2.17 In Figure S above We estimate that the proponiorr of total ICANN expenditure
speciiically on, or in support of the GNSO increased from around 2 per cent in 2003-04
to around 6 per cent in 2005-06, With a provisional increase of another one percentage
point for the fiscal year 2006-07. We Were not able to obtain exact figures for ICANN
expenditure on the GNSO, and so these estimates are based on staff costs, overheads
and other related expenditure (see Annex A. Figure A41 for details). The increase in
expenditure is largely accounted for by salary and overhead costs for additional ICANN
staff support. Although the benefits from Lhjs increased expenditure are realized in
terms of stronger executive policy coordination, the Constituencies have not received
incremental Binds from ICANN over the years to help build administrative and outreach
activity.
29

LSE Public Policy Group                                                                GNSO Review
2.13 For Constituencies, their Website is (or should be) one of the pximaly interthces
With potential new membexs or participants in the policy process Who are not already
involved. We found considerable vaxiation in the quzhty and content of Constituency
Websites (see Figure 9 below). Some Constituency Websites Were lamentably out-of-
dzte or lacking in basic
Figure 9: Analysis of Constituency websites and the presence of 50 desirable
feature
TM mama team;
so
otto Re| my
xv
Ncu
Ra|m`
isp
o           to          20          30          40          So
Number offeatures found (out of 50)
mMmtmh=p     I patty ampm    I cmtm    a Hetpotnanmunnma EXPIMM
No-rn The g-rto Regmnes have setfyaesnea membashtp and thaefm Wnum named tm use that Webstte tm entnurage new
members WE Wnum themfm expect amtaavety laws` Mm an the mambashtp and tamed tategnry
same Lse ppo analysts Df cnnsummty Webmes
details on how to join or any explanation of the potential benefits from joining. The
Registrar Constituency currently has two separate Websites, WWW.icann-registmrsogg
signposted from the ICANN Website and answer version hidden away at
WWW.icanmegg~straIs.org. The ISP Website ceasedto timction duting this research, and
still refers to the Names Council, suggesting that it has not been updated since early
2003. We canied out an `unobtmsive measures' analysis of Constituency Websites. We
compiled a list of around 50 possible features that one might reasonably expect
Constituency Websites to have. This is not an exhaustive ljst, neither is it based on any
detinitive list of features Websites should have. The list of questions is reproduced in
Annex Figure A24. We coded each Constituency Website for each of these 50 items to
30

LSE Public Policy Group                                                           GNSO Review
get a sense of the diversity ofinformation available. The `Total' row bar shows the
maximum number of coded features per categoxy, and the bals show how
Constituencies actually pexformed by compaxison. The quality of Constituency Websites
vaxied greatly, With five out of six Constituencies exhibiting less than two Iitths of our
coded features. By far the most comprehensive Website belonged to the BC, With just
under four Jitths of coded features.
2.19 We asked Constituency member organizations in our on]jne survey to score
different aspects oftheir own Constituency Websites. We counted the number of times
each aspect mnked highest or lowest in each response, shown in Figure 10. Here
positive numbers in the net score column show overa].1 satisfaction and negative scores
show dissatisfaction. Cument members tend to be most satistied With the Way in Which
Constituency Websites do quite stmightforward administrative functions, such as
making reports and minutes available and posting contact details for oftice holdexs.
Respondents tend to be less positive about more sophisticated functions, such as
tbcilitating open discussion and especially attracting new membexs. The strong
consensus amongst our interviewees is that Constituency Websites currently do vexy
little to attract new members to the GNSO.
Figure 10: Survey respondens' views on different aspects of their Constituency
website
Net score
Ranked     Ranked          .
highest        lowest          (highest 7
lowest
Post~ ; contact details of oftice holders                  26            5               21
| re |oxts and minutes available                    21 |        13
Keeping up-to-date With relevant GNSO news
and issues                                                                                          2 1                                                  13
Kee | | members' contact details u|-to-date             20 |         11
Facilitat  ; o|en discussion on ke  issues                  21              13 |
Attractin| new members to the Constituenc |       24                -15
same Lse ppa Drama sway Dfcnnsumenty members
2.20 There are currently considerable vaxiations in the constitutional and opemtional
axrangements across Constituencies. These differences mise the transaction costs of
paxticipation for organizations and individuals Who might Want to paxticipate in
ICANN's Work. especially if they are interested in more than one Constituency. The
Constituency Websites are divexse in terms oftheir bmndtng and content and may
3 1

LSE Public Policy Group                                                                GNSO Review
appear confusi.ng to potential new members. In some cases Websites are dormant even
for existing membexs. This is an example Where divexsity can undermines
representation rather than encoumging it. We do not believe that relying on
Constituencies to improve this situation on their own will be feasible (particularly given
the varying levels of scarce resources). There are currently some impoxtant cross-
Constituency groups and meetings, but these genemlly involve clustexs of specific
Constituencies rather than regulaxized and inclusive interaction across all Constituency
executives.
2.21 We recognize that the Constituencies are by nature divexse and rely on a cextain
level of autonomy in terms of establishing procedures and positions. Nevextheless it is
our view that much can be gained from a more coherent approach to managing and
growing shared functions and capabilities across the Constituencies. Despite obvious
diversity, Constituencies share important administmtive and strategic functions. A more
standardized approach to supporting constituencies from ICANN staffhas the potential
to lower costs for Constituencies and to greatly increase their accessibility for members,
and potential member (and non-member) organizations.
Recommendation 3
There need: to be greater coherence and standardization across Constituency
operation:.  For this to work efectfvely, more JCANN slgysupporl would be neededfor
constituencies.
2.22 A more `off-the-shelf` approach by ICANN to supporting Constituencies might
involve making available to Constituencies some pre-packaged products that they could
apply to their own administmtion and membexship-attracting stmtegies. For example,
this might involve consolidation of Constituency Websites as sub-parts of the ICANN
Website in such a Way that each Constituency has a section that is coherently branded
but Which allows them the flexibility to publish and communicate in their own Way.
Each Constituency might have direct content management access to their respective
sub-sites. Mailing lists and other comespondence should be secure to Constituency
membexs only. Although Constituencies are inherently different, there is great potential
for them to develop, With ICANN's help, common products to facilitate administration
and outreach at Constituency level. This might include design ofgenexic ICANN
marketing literature Which can be tailored by Constituencies according to their own
32

LSE Public Policy Group                                                                GNSO Review
needs. We reviewed Constituency statements in the paragmphs above, and there is
certainly opportunity to develop more standardized formatting for statements based
around issue analysis for different pieces ofpolicy. Generic ICANN email alerts or RSS
feeds might also be developed from this more coherent platform, allowing
Constituencies to offer members ICANN news updates and latest policy Work. The
essential objective here is that the link between Constituencies, the GNSO, and ICANN
should be far more obvious and understandable to potential members (and the Wider
Intemet community).
Recommendation 4
A GNSO Constituency support omcer should be appointed to he? Constituencies
develop their opemiiomg websiles and oulreach uclfvily.
2.23 Following on from Recommendation 3, it is important to consider how best to
organize the resources for developing an `off-the-shelf` approach to supporting the
needs of Constituencies, at minimal cost. ICANN policy and opemtiortal staff could
institute some kind of forum for developing and sharing professional tools and pmctices
across Constituencies. But there also seems a strong mtiortale for strengthening the
support available to Constituencies by creating a GNSO Constituency Ofticer, With a
key role of supporting common corporate development Work across the Constituencies.
How Well do the current Constituencies represent the diversity of
stakeholders in their areas of interest?
2.24 In evaluating the representativeness of Constituencies We also sought to look at the
extent to Which the mnge of current Constituency member organizations reflects the
global distribution of domain names registered under g'I'LDs and Irttemet usage in
general. We do not believe that it Would be ameaningiijl exercise to try to establish the
extent to Which Constituency member organizations are actuz].ly statistically
representative of the global population of individuals or organizations, particularly not
in relation to the global commercial and third sectors. This is because ICANN's role in
the govemance of the Irttemet is arelatively specitic one. At any one time only a small
(and inherently hard to establish) proportion of civil society organizations will be
interested in or concerned about its policy decisions and detailed debates on Irttemet
names and numbers. However it is possible to evaluate very broadly to What extent
33

LSE Public Policy Group                                                                GNSO Review
cument membexship broadly reflects global diversity. We looked at Constituency
member lists, allocated each member to a global region, and then calculated the
maximum number of Constituency votes that could be cast by organizations in each
region. Where infoxmaiion Was available, We have used specijic Constituency voting
allocations to do Lhis. A disttibution of voting power for each Constituency can be
constructed for each global region (assuming that all Constituency members vote). We
Figure 11: Estimated global distribution of Constituency membership
|            Approximate ICANN global regions               Categoxized
.               .                                   .                   as
Europe      Afnca       Asia        Nonh          Latm                     .
.                  .                  .          Intemahonal
Pacljic     Amenca     Amenca
and
Caxibbean
Estimate d
proponion of
domain names
.          .               19            0.2           11                                 4                   Na
registered ].n
GTLD space
| 0  rom!
Estimate d
percentage of
Imemet users             29                3               33                23                   7                       Na
Worldwide (%
0  rom!
The following Iigures give an indication of the percentage disttibution ofmaximum
possible votes across ICANN global regions for each Constituency.
RH|      42  =_IZ  58 |
|     23  _*  15    55     2 |
BC                          26             3              4              35 |           31
|  25  _*  7    35     2      30
NCU                        12 |       zs             so               11                    12
Isp                         41             3             lo |         lo                   21
NOTE I We coded each consmumcy member by as allotted number ofvotes aLcOnSL=mmCyleve1,and the global
regwn m Wmh =L `S regmered ms gwes a very genem1=nd=caL=On ofme extenuo Whmh consnmenaes are over Or
undemepresented m dmrent global regwns
NOTE z The data m am we `S based On the ICANN regwns Conmmed at the ICANN annual meenng m Mamma =n
june zoo; gm NWWW `cam Mg/montw/geOYreg=OnsYwE=c mm) ms cawgonzanon puts counmes m the Mule
East and the caucasus =n the Am pause global regwn We are not aware ofany mnher modmcanons :0 ICANN
global regwns Smcejune zoo;
NOTE 3 The Be asked us :0 Categonze One ofum members based m jordan m the MM regm pamcularxy as
much ofthe acnvxty Owns Orgamzanon `S Mated m North Amca The data ah eve reflects am request
SOME LSE pm and as Df = ma `mama. data and cnnsmuem  membashl| data
34

LSE Public Policy Group                                                           GNSO Review
then calculated the proportions of `total domain names registered under gTLDs' and
`total Intemet usexs' for each region. Figure 11 above compares the Constituency vote
disttibution by global region With these two indicatoxs of disttibution of global Intemet
activity. It is Woxth mentioning that the data in Figure 11 is based on the ICANN global
regions, Which have not been subject to change since 2000. Turkey and countries in the
Middle East are currently categoxized as Asia Pacitic (perhaps a slight anomaly), and it
may be Woxth the GNSO and ICANN as awhole Working to retina global region
detinitions to capture changing patterns in Intemet usage in these regions. Also,
representative organizations are otten international in focus and therefore cannot be
eadly categoxized by global region. We have established an Intemational categoxy for
membexs, Which have a global spread (most commonly in the ]P and BC). Although
Constituencies provide guideljnes for voting rights for Constituency membexs, We
found it at times difticult to establish how many votes member organizations actuz].ly
received.
2.25 A tixst rough indication of the proportionahty of Constituency membership in
Figure 11 is in texms of global disttibution ofgTLD registmtions. There is aheavy skew
here towards domain name registrations in Noxth Ameiica (two thirds of the total), and
to assist compaxison the Constituencies are ordered by the representation ofNoIth
Amexica users. By compaxison With the proportion of United States gTLD registmtions
most constituencies are lower, While g'I'LD Registties and Registmrs are somewhat
closer to the Noxth Amexica percentage tigure. The NCU Constituency has some
indication of over-representation stakeholdexs in three out of live global regions, Which
of coutse is no bad thing. This suggests that divexsity of membership in the NCU is
relatively strong. The Registmr Constituency is also relatively propoxtional across all
live regions. The BC and ]P Constituency are interesting as they both show relatively
similar disttibutions across the live regions and pattems Which are quite close to the
overall disttibution ofregistrations, suggesting a balance of membership power close to
the actual global disttibution of gTLDs. The ISP Constituency has a strong over-
representation i.n Europe. The ISP Constituency has organized a global meeting for
ISPs.
2.26 However, the second row of Figure 11 shows an altemative compamtor, an
estimated disttibution for the number of global Intemet usexs (at the end of 2005). This
35

LSE Public Policy Group                                                                GNSO Review
gives avery different picture, With much less of a skew towards North America, and a
much stronger showing in Europe and Asia Pacific. Whereas two thirds of all g'I'LD
registmtions are in North America, just under one quarter of Intemet users are located in
this region. In terms of objectives for Constituencies, particularly user Constituencies,
the user distribution might be closer to the one Which Constituencies could aspire to
approximate in terms of growing their own membership. The NCU Constituency shows
quite a close it to the distribution of global Intemet users across at least four out of live
regions. The BC and IP Constituencies show under-representation in two out of live
regions, but it is important to take into account that around one thjrd of the members for
each of these Constituencies are hltemationahy-oriented representative bodies.
2.27 We looked for away of evaluating the size and profile of Constituency member
organizations, particularly the organizations represented by the most active individuals
in each Constituency. We noted above an apparent rule of thumb that around one
quarter of Constituency members are responsible for around three quarters of the
activity. Using mailing lists or meeting attendance as indicators of activity, We ordered
the most active participants for each Constituency since early 2002, recorded the
organizations they represented, and focused on the Top Ten most active organizations
for each ofthe four user Constituencies - BC, ]P, ISP and NCU. We collected Alexa
metrics data for these active member organizations' Website homepages, including
`reach per one million users' and `tmfiic ranking of the Website'. For each Constituency
We looked at the top-ranking member organization and the median member
organization for each of these two indicators. Figure 12 below shows the results for
each of the four user Constituencies.
36

LSE Public Policy Group                                                                GNSO Review
Figure 12: Estimating the web profiles of the top ten most active member
organizations in four user Constituencies
|    Reach per million users        Traffic rank of Website     Number of
.                                          .               organizations
Top                Median           Top             Median
too sma].1 to
member          member          member       member          In
eature
BC               13,135        21          26      62,284          2
|    2]     536 |
NCU                128         3        12,131     290,960         1
Isp                                                      Insumczenmuzu amiable
Note: This table gives an indication ofthe web projile ofthose member organizations that are most active at
Constituency level. We have reviewed available data on the Top 1o most actively represented organizations in
each Constituency since June zooz. We have used www.alexa.com to get an indicative tmdic rank and `reach
per million users' profile for each mganization. The data shows the top and the median score for each
Constituency.
same LSE ppo and as Dmsxa Webm Ms: and Webstte user data
2.23 Looking in more detail Iitst at the Business Constituency (BC) the mix oftop ten
most active organizations appeals to be somewhat miscellaneous. Not all are household
names, and although one or two big bmnd names appear, they tend to be towards the
bottom of the table and have largely IP based interests. The top milking member
organization in texms of `reach per million usexs' is the Wa1tDisney Company, With on
avemge two in every 100 people visiting this Website (the 26Lh highest ranked Website in
terms of ovem].1 traffic): however, representatives from the Wa1tDisney Company have
attended six meetings out of 24 since 2002, and infrequently in the last two years,
suggesting `distanced' participation at best. Big name organizations such as Vexizon
and the News Coxpomtion give the Top 10 commercial credibility and explain Why the
median scores in Figure 11 for the Business Constituency are on a par With the IP
Constituency.
2.29 Our evidence clearly shows that there is a committed core ofBC representatives
some ofwhom are affiliated to large and Well-known commercial coxporations and
tmde associations. However, data (in Annex A. Figures A15 and AIS), suggest that
direct paxticipation rates tend to fa].1 off relatively quickly. Although around 33
representatives ofBC member organizations have attended at least one BC meeting
since 2002, around 25 representatives have attended 5 meetings or less (out of a
possible total of 24). We estimate that 5 member organizations have attended more than
37

LSE Public Policy Group                                                                GNSO Review
10 of the 24 meetings for Which minutes Were available. Since 2002 only the three
cument Constituency representatives have attended at least three quaxters of all meetings
for Which minutes Were available. Three mqjor coxpomtions and four representative
associations listed as Constituency members have never attended aBC meeting as for as
our analysis shows.
2.30 Looking at the type of organizations most active in the BC, there are three
representative organizations for businesses, two mqjor global telecommunications
companies, two mqj or global content provider companies, a leading Middle Eastem
accountancy and IP Iitm, and a range of small single-pexson consultancy businesses
(see Annex A, Figure A27). Looking at the BC membexship list in 6]].1, there is a much
Wider spread of organizations, including some Well-known Intemet-specific names like
Yahoo, Inc. It is vexy hard to establish to What extent these organizations actuz].ly play a
pan in the GNSO process. Some intexviewees argued to us that the BC should aim to
grow its coverage of new Intemet-related markets, particularly e-commerce and Web
hosting companies, and encoumge new organizations to participate actively in policy
development. Our intexviews With technology business associations familiar With the
BC echoed this point, suggesting that BC does not currently provide ahome for new
types of organizations in the World of e-commerce. Other intexviewees suggested that
other mqjor lntemet-related business sectors such as Iinancial services and retail
industxies did not seem to figure on the cument membership list. Figure 13 below gives
anindication of the level of direct penetration of the current membership into some key
business se ctors.
33

LSE Public Policy Group                                                                GNSO Review
Figure 13: The market value of oorporations who are members of the Business and
Intellectual Property Constituencies
Total cumulative market    Number or Bc ma n)       Market Mm or
Mm or ute Top zs             Constituency          utese Constituency
campuses in 'ms sector    members alnongst ute         members as 8
Sector                                          (sum)                           Top zs              percentage or Top zs
sector
market value
Bmg |
Telec ammunicatians                        1,237                               4                                2
Diversified Fimmcials                      1,115 |
Hardware and equipment                 1,oso                               2 |
Software services                              $31                                1 |
Food, Drink and Tobacco                 gso                             1                             7
MM         926 |
Mling |
Media                                              607                               3                              31
Conglomerates                                  575 |
some; Forbes Global zooo
2.31 Tuming to the JP Constituency, the ten most active member organizations tend to
be representative associations. These organizations have large and Well-known
coxporate clients and membexs, and so a similar Website analysis can be slightly
misleading, because Websites of the representative organizations may not always be as
Well visited as the Websites oftheir membexs. The median figures for the IP
Constituency in Figure 12 suggest a similar profile to the BC Constituency in terms of
usexs per million Intemet usexs and traftic volumes. However, perhaps not surpxisingly,
the largest member organization in this Constituency has a much smaller Web projile
than the largest one in the BC. It is interesting that all bar none of the most active
member organizations in the ]P Constituency have arecognizable and active Website (a
possible sign that the ten most active member organizations in the ]P Constituency are
all Well established enough to have a fimctionjng Website).
2.32 Our data on the IP Constituency tends to conlixm a picture of relatively solid
paxticipation by some mqj or global IP stakeholders. Detailed data (at Annex A. Figures
39

LSE Public Policy Group                                                                GNSO Review
AZS) show that at least 17 out of 20 mqj or intemational representative bodies have
attended at least one IP Constituency meeting. However, there is still quite a high
proportion of Constituency members in the categoxies of regional and national members
that have never attended a Constituency meeting. Again the picture is one of a strong
and relatively small core of active participants. The top 12 participant organizations
include some mqj or names, both representative bodies and coxpomtions. Evidence on
paxticipation elsewhere in this repoxt shows that the IP Constituency Council
representatives are perhaps less active than other Constituencies at Council level, and
have shown quite high rate of Councilor tumover (see Annex Figures A17 and A45).
Despite numerous remjndexs to the Constituency members about our on.lj.ne survey, We
received sutpxisingly few retums given the apparently reasonable levels of activity by
Constituency membexs that our data indicates.
2.33 For the ISP Constituency We Were unable to establish atop ten most active
organizations, due to lack of data and insufficient smvey responses. However, the ISPC
have told us that that the level ofpaxticipation in the last live ICANN meetings
avemged over 20 individuals With more than 14 different organizations represented, a
point bome out in obsexvation of an ISPCP session at the ICANN annual meeting in
Wellington.
2.34 The NCU Constituency appeals to be quite diversely representative at least in
terms of distribution of its members. It nevertheless scores comparatively lower in the
size and protile of its most active members in compaxison to the BC and the IP
Constituencies. This is perhaps understandable considexing the organizations involved
in these three Constituencies are vexy different. The Websites of the most active NCUC
membexs tend to be signiticantly less visited and have signiticantly less incoming
tmftic, although only one active member organization has a Website Which is not
recognizable by Alexa. These indicators suggest therefore that the NCUC tends to be
diverse yet relatively small-scale in its membership profile. The vexy limited resources
available to this Constituency constmin the extent to Which outreach Work can take
place.
2.35 Although the NCU Constituency membership is quite divexse When looked at from
Within, it is largely meaningless to txy to come to any statistical conclusion about how
40

LSE Public Policy Group                                                                GNSO Review
reflective the current membexship is of the Wider population of global non-commercial
organizations. One Way forward for the NCU Constituency however might be to
identify some key indicatoxs of spread against Which its membexship can be evaluated.
We suggested above that global Intemet users and registered domain names in the
gTLD space could be two genexic indicators of balance. More specijic indicalors for the
NCU Constituency are entirely feasible, such as density ofnon-governmental
organizations by global region and percentage of organizations opemiing in different
policy areas. Figure 14 gives an indication of density ofnon-governmental
organizations by global region. It is interesting here to see that NGOs are much better
developed in higher income than lower income economies, although the gap is less in
Latin Amexica. Looking at the vote share for the NCUC, Latin Amexica comes fourth
out of live in terms of global spread, and so We might conclude, in very tentative texms
of coutse, that the NCUC could look to grow its presence in Latin Amexican non-
govemmental networks. Figure 15 shows the disttibution ofNGOs activity across
different types of activity. Again this might sexve as a function-specijic guide for
growing NCU Constituency activity.
Figure 14: The density of nomgovernmental organizations by global re|on and
level of economic development
Global Civil Society
Low and middle          High income          Total
income economies       economies
Euro|e / Middle East                   1.9              34.3         16.1
Nonh Amelica |          13.0                 13.0
mm AmeIica/ cmbbean |         zo.3                 1o.o
Asia Pacific                                                3.5                          12.0                  5.1
Aim                  1.5 |  1.5
NOTE: Figures in this table show density by `number ofNGOs per million ofpopulation'. Fm each
global region we have calculated average density.
alms LSE pm Max S`: O dam am azabaz awzsmg: yewbwk 2004/as
41

LSE Public Policy Group                                                           GNSO Review
Figure 15: The average percentage of activity by nomgovernmental organizations
across different areas of activity by the level of economic development
Global Civil Society
Low and middle         High income          Total
NGQs acme In                                 income economies        economies
Economic development,                      25.3                  23.6             24.6
infmstmcture
Research                                            20.2                     22.1               21.0
Law, |o]jc  and advocac                    12.6                 11.6            12.3
Social develo|   ent                               12.1                     10.3               11.5
Education |
NOTE: Figures in this table show the average percentage of total NGO activity relating to different areas.
slung LSE Pm am! ns O dam am azalea! awzsmg: yembwk 2004/os
2.36 The cument system of GNSO's opemtions is not very old, stemming from a
decision made in 2002 to separate out the Supporting Organizations for genexic names
(gTLDs) and for countxy code names (ccTLDs). Pxior to this decision there Was a single
Domain Name Supporting Organization (DNSO), Where countty code names Were
represented by one of seven Constituencies (the other six effectively making up the
cument Constituencies in the GNSO). When the GNSO Was established in December
2002 as a sepamte body dealing speciiicahy With gTLD policy development, the six
Constituencies Were largely canied over. Yet even in the course of a few years, there
are mixed signs of growth and atrophy in the quzhty of representation and the degree of
actual participation across the Constituencies.
2.37 Some aspects of the cument GNSO Constituency system seem to reflect a snapshot
of the interest groupings most active on genexic names issues in the founding stages of
ICANN in the late 1990s. Many intexviewees told us that the oxigins of the current
Constituency system are found in the organized interests that played key roles in
establishing ICANN at that period. These included mqor Ameiican computer and
telecommunications corpomtions (such as AT&T, MCI, IBM and Te1stra)Who Were
engaged With the process; some mqjor content provider corporations (such as Disney
and Time Wamer); representative associations in the intellectual property area; and
early-established elements of the registmtion industry (badcally Network Solutions).
Intemet service providers (ISPs) Were also represented since they Were playing a mqor
role in helping consumers to gain access to domains and URLs. Constituencies at this
42

LSE Public Policy Group                                                                GNSO Review
time Were `self-organizing' (as stated in the GNSO Bylaws), Which some interviewees
told us meant the `loudest voices prevailed'.
2.33 A fimher piece of the snapshot involved the presence ofintellectuzl property
interests. Many of our interviewees agreed that by the end of the 1990s intellectual
property issues relating to the use of domain names Were in the late stages of
formalization With the development of the Uniform Domain Name Resolution Process
(UDRP). The World Intellectual Property Organization (WIPO) held a public hearing in
the period between the 1993 White Paper and the formal recognition of ICANN in early
1999, and issued a report suggesting features of the UDRP. The rules eventually
adopted by ICANN differed from those in certain respects. ICANN's Board and a
`small dmhing group' composed of various ICANN Constituency representatives
dmhed the UDRP in use today. Analysis of some of the mqj or public forum discussions
shows that the original UDRP adoption generated the largest amount of discussion (in
terms of numbers of corttributors and Words Written) of any subsequent policy
development process discussion. Subsequently the UDRP has come to be Widely
regarded as a successful initiative, and agood cross-section of our interviewees
suggested in varying degrees that it could be viewed as one of ICANN's successes.
However, some critics and researchers argue that it is biased towards complairtartts.
This necessarily brief historical background in the previous pamgraphs brings out the
importance of factoring in the context under Which the GNSO Constituencies Were
established. This is particularly the case When considering the current picture of
representation and participation in the GNSO and thinki.ng about feasible participation
conditions for the future.
Recommendation 5
Constituencies shouldfocus on growing balanced representation and active
purlicxpaiion broadly proportional to wider global distribution: for relevant indicator:.
2.39 It Would not be usefhl to recommend any specific targets or thresholds for
membership or participation, Which Would be alargely meaningless exercise. However,
there are anumber of possible examples of global indicators that Constituencies could
use to guide future Work on outreach and effort to grow membership and participation.
These may be very general such as the percentage of domain names registered under
gTLDs, or they may be specitic to particular Constituencies, such as the global
43

LSE Public Policy Group                                                                GNSO Review
disttibution of non-governmental organizations Working in the Held of new
technologies. Growth in membership should be monitored by Constituencies to ensure
that distribution is as balanced and propoxtional as possible in relation to key indicatoxs.
2.40 In response to questions about representation and paxticipation, many intexviewees
argued to us that although the secutity and stability of the Intemet is of vital impoxtance
to many organizations, it is unrealistic to expect people from these organizations to
Want to continuously participate in developing policy on it. As one person shrewdly put
it, `most people Want to use the Intemet, not develop policy on it'. Although this view
seems intuitively compelling, there are currently policy issues on the horizon Which are
directly relevant and interesting to a Wide range of actoxs, notably data pxivacy and
]DNs. Individuals or organizations Who find they have lost a valuable domain name to
fraudulent or overly exubemnt resale activity have a direct interest in institutional
structures that allow them to influence policy (or at least have a say on policy issues).
Government organizations have also shown themselves increasingly interested in Work
that the GNSO is doing on issues relating to Website registmtion data. These are two
examples of how technical and otien complicated GNSO policy issues tmnslate into
actual, real-life issues for a Wide mnge ofpeople. So We believe that there is scope to
improve the current level of involvement in the GNSO's Work.
2.41 In our survey completed by non-member organizations (that is, any organization
not currently a member of a GNSO Constituency) We asked respondents to score thctors
Which Would be impoxtant to them and Would encourage them to become amember of
the relevant GNSO Constituency. We also asked respondents about potential baniers
that Would explain Why they are not currently members of the relevant Constituency.
Figure 16 (below) summaxizes the most popular responses and suggests two impoxtant
points. Fitst, While it conlitms that ensuring the stability and secutity of the Intemet is a
vitally important consideration for organizations, it illustmtes that another important
motivation involves helping to shape the fhture of the Intemet. So the view put forward
in the previous pamgmph (that vexy few organizations actually Want to make policy on
gTLDs) is perhaps too pessimistic. A second, closely related point is that the main
baniers that organizations identiiied are to do With feeling alienated or distanced from
the ICANN process, not really knowing how to participate, or simply not being aware
ofthe GNSO process. In fhct, the least cited reason for notjoinjng GNSO Was that
44

LSE Public Policy Group                                                           GNSO Review
`genexic domain name policy is not a pximaly concem for our organization', suggesting
apositive potential to grow the membership of GNSO Constituencies.
Figure 16: Reasons |'ven by survey respondents for why their organization would
consider joining a GNSO Constituency, and why their organization is not
currently a member of the relevant GNSO Constituency
|            HIGHEST                       zn| HIGHEST
Reason to get             Ens-ming secutity and            Helping to shape the future of
Involved                   stability of the Intemet           the Intemet
Reason not to get         It is difticult for an                 We Were not aware of the
Involved                   organization ]j.ke ours to         GNSO Constituency system
influence these kinds of
Issues
NOTE: Annex A Figures A30 and A31 show the full results.
same LSE ppa mama gm| Dfnmmembernr|mtzeanns
2.42 Our interviews and research into other mqjor international representative bodies
such as the International Chamber of Commerce (ICC) or the World Information
Technology Sexvices Association (WITSA) suggest that the entry level for membexship
can be a decisive tbctor in encouraging awide vaxiety of organizations to participate in
policy development. In both these comparators, tixms and bodies become members at
the corporate or `Whole organization' level, and are then tiltered towards the relevant
policy fomms or commissions. For example, organizations interested in paxticipating in
the Work of the ICC join the ICC as local or national level membexs, and are channeled
towards the relevant Commissions or special Working groups. Similarly, organizations
canjoin the ITU and then paxticipate in one of the three Sectoxs according to their own
interests. We found little sense that `authentic' bottom-up processes ofpaxticipation are
undermined by having this Whole-organization route in. In our discussions With
representatives from a mnge of representative associations (see Annex C), intexviewees
suggested that this system provides arange of obvious benefits: in paxticular a more
professionahzed support environment for member organizations; greater flexibility for
membexs to pick and choose policy discussions; and lower administrative costs for the
member organization itself in terms of communicating With members and keeping them
aware of ongoing Work and so on.
45

LSE Public Policy Group                                                                GNSO Review
2.43 Currently the main intexface With the GNSO is solely through becoming a member
of a relevant Constituency. Many intexviewees pointed out to us that joining
existing Constituencies that (lightly or Wrongly) appear to be subject to control by
relatively small groups of interests is otten vexy off-putting to prospective membexs. If
they react adversely to perceiving such defects in Constituencies, potential membexs
may then steer clear of any involvement With GNSO and ICANN as a Whole. There are
also dissatisfactions about the vaxiation in pmctices across Constituencies and the
apparently resttictive and now dated missions of current Constituencies. Resources at
the Constituency level are otten limited and lack standardization | a consequence of the
bottom-up stakeholder philosophy, but also a tbctor that considembly increases the
perceived complexity and impenetrability of GNSO's opemtions. The activities of
Constituency executives and their levels of outreach activity and membexship suppoxt
have been vaxiable and are hard to predict. Finally having to take out membexship only
via an amateur-run Constituency gives the impression to potential members of being far
removed from the heart of ICANN and from influence on its policy processes.
Consequently the marginal benefits ofbeing a stakeholder in the ICANN process may
seem diffijse.
Recommendation 6
The ba.vi.vforparIic?a1i<m in GNSO aclfvilies need: to be revised from Constituency-
bmed members)'? to one deriving from direct JCANN smkeholderparlicwaiion.
2.44 A key change that could help boost GNSO membership Would be for organizations
and individuals to become stakeholders in ICANN through a centmlized stakeholder
process. The pximaly intertbce Would be With ICANN rather than With GNSO
Constituencies. And a more standardized and coherent ICANN stakeholder package
should be developed so as to allow new organizations to feel much more directly
involved. We also show below (pamgmphs 3.2 and 3.3) that ICANN is much more
visible to the Intemet community as awhole than is GNSO, so that for users especially
the information costs ofj oining ICANN Will be much less than those involved in
Ending out about, understanding andjoining the current GNSO Constituencies.
Depending on their interests and legal status, new ICANN membexs Would be
channeled towards the relevant stakeholder Constituencies (or to other representative
channels if member interests lay in non-gTLD issues). Membership fees could be tiered
to individuals and to the size and type of organizations or Iitms (as is currently the case
46

LSE Public Policy Group                                                           GNSO Review
With some Constituencies), but a single scale Would prevail. Revenues from fees Would
be centrally collected by ICANN. Expenditure on Constituency professional Work and
outreach development Would come from a remitted mqj or element of ICANN
stakeholder fees. And pexforming successfully (such as attracting ahigh rate of
potential membexs as actual members, or improving panicipaiion levels appreciably)
should also attract a commensurate conttibution from the centml ICANN budget, so as
to give (reformed) Constituencies incentives to be proactive in attracting membexs and
conttibutors.
47

LSE Public Policy Group                                                           GNSO Review
Part 3:
The sdsibility, transparency and openness of the GNSO
process
3.1 The transparency criterion is cortcemed primarily With the GNSO process being
open to inspection and scrutiny by internal and external actors. If a decision-making
process is to be transparent, it must first of all be easily visible and accessible.
Potentially interested stakeholders must be able to find out What is going on, Which
organizations are involved, What timetables and policy processes are in opemtion, What
kinds of considemtiort are being looked at, and how they themselves can best fo].loW
and influence discussiorts. A key question is therefore Whether the current opemtions of
GNSO create sufticiertt visibility to allow the essential foundation stone of transparency
to be seen to be in place?
The external visibility of the GNSO
3.2 The opemtions of ICANN have been Well described in artumber of recent extensive
and detailed published accounts by academics or impartial observers. However, in all
such accounts the Generic Names Supporting Organization has very low visibility
compared to that of ICANN as a Whole. For instance, in a414 page report on ICANN
Written by the National Academy of Sciences in the United States there is a detailed
contents page running to 7 pages, Which does not include any mention of the GNSO.
Similarly, a search in Google Books generates 3 hits on the GNSO compared to around
1,600 for ICANN. Tuming to the coverage of ICANN on the Web and in Intemet
mailing lists, We have found only a very tiny proportion of references to ICANN Which
make any reference at all to GNSO as awhole or to the GNSO Council. A basic search
for the terms `ICANN' and `GNSO' in mqor international Written media databases
such as Nexus Lexis Executive shows that GNSO results tend to be signiticarttly less
than 1 per cent of equivalent results for ICANN. Searching the IETF Website, We tind
around SO results for the GNSO compared to just over 34,000 for ICANN (see Annex
A. Figure A32 for some indicative search results).
43

LSE Public Policy Group                                                           GNSO Review
Figure 17: Comparing the web visibility of the GNSO and ICANN
-       Search for             Search for
`GNSO'            `ICANN'
Average number of results across four different                                         _  _
search engines (Google, Yahoo, Almwsm, and            1507000          13 ml]-1-Ion
]\/IS`
Estimated percentage of total results in. ..
mm gTLD |         53
.org g'I'LD                                                                 71                    21
(Ofwhich were icw'm.org2                 (45)            (49
courtly code TLDs |           16
Other gTLDs                                                             12                     10
Total percentage                                                                              100                           100
same LSE ppa may5,5 nfsemhmguus sum fm Estahhshea `memo search Engmes
3.3 A fhrther insight into the relative visibility of ICANN compared to GNSO is the
scale of their two Web presences given in Figure 17 above. In order to get a rough
picture of Where the GNSO Web presence Was strongest (i.e. in terms of TLDs), We mn
searches for the terms `ICANN' and `GNSO' using four established search engines,
restricting the searches to individual generic TLDs and a]1 country TLDs. We took the
estimated total number of results for each search, across each search engine, and
calculated an average across the four. The avemge number of search results for
`ICANN' Was around 13 million compared to 150,000 for `GNSO' (about 1 per cent).
An interesting contmst also appears in the ]j.nes below. Around half of the results for
`ICANN' appear in the dot.com domain (including press, commercial sector, and other
private sites), compared to S per cent for `GNSO'. Perhaps not surprisingly, the great
proportion of results for `GNSO' appear in the dot.org domain. However, using the
Google page ]j.n.k fimction, We found that around 50 per cent of]j.n.ks to the GNSO
homepage originate from Within the icartrtorg domain, With a fhrther 43 per cent
originating from one Website WWW.1atinamericann.org a Latin American NGO With a
]j.nk to the GNSO on every page ofits Website. These results are suggestive of the
extent to Which current knowledge of and commentary about GNSO is confined to
`deep insiders'.
49

LSE Public Policy Group                                                                GNSO Review
3.4 At the outset of our Work for this Review We set up a multilinguzl research Website
at WWW.ican.n-gggorevieW.ogg in order to canvas as many views on the GNSO as
possible from around the World. The research Was advertised in the global press, and We
employed a group often intemational post-graduate students over apexiod of six Weeks
to identify key organizations from the commercial and civil society, contact relevant
individuals Within these organizations, and invite them to comment on the GNSO or
more generahy on genexic TLD policy development at ICANN. The response to this
intensive Work to build awareness for the Review Was extremely disappointing. We
received avexy limited response, suggesting to us that the range of opinions about, and
the level of awareness of the GNSO outside of the already active ICANN community
itself Was vexy low. Our discussions With membexs ofgovemment agencies in the USA,
UK, the EU, the OECD and With members of the Govemment Advisory Committee
(GAC) also revealed relatively low awareness amongst ofticials directly involved With
ICANN ofthe Work ofthe GNSO. Our interviewees assured us that We should expect
general levels of awareness of GNSO in the countty governments themselves and in
mqjor intemational organizations to be even lower. This view Was also partially
coniitmed by several senior government officials With extensive involvement in
Intemet issues Whom We approached for intexviews, Who declined because they said
that they did not know enough about GNSO's operations to comment.
3.5 Many insiders have argued to us that `GNSO is ICANN', in two impoxtant senses.
First, looking at revenue data and other outputs, GNSO issues, or at least issues relating
to GNSO stakeholders, are at the healt ofthe ICANN process. Second, GNSO is in
many Ways the most distinctive and most established piece of machjnexy supporting
ICANN's claim to facilitate `bottom-up, stakeholder participation' in an open and
tmnsparent Way in the govemance of the lntemet. The GNSO is therefore a strange
combination of being a cxitical and integml pan of ICANN, but one that is almost
invisible to people Who are not already closely involved With ICANN.
3.6 A main reason Why GNSO's visibility on the Intemet is cumently very low is that
there have been sexious deticiencies in the design of ICANN's overall Website over
recent yeals. These problems have long been known to the Board and the ICANN Chief
Executive, axising from previous restxictions on resources and paxticular pexsormel
50

LSE Public Policy Group                                                           GNSO Review
issues. However, Website problems in the modem age cannot be treated as pexipheral or
as involving only a dispensable or luxury good. Especially for abody such as ICANN
and its main components such as GNSO, aproperly Working and designed Website is an
integml element ofbeing an effective organization at all, and its role in respect of
tbcilitating transparency is of cxitical impoxtance. Our intexviewees almost univexsally
expressed deep fmstmtion With the cument Web provision. Respondents to our suxvey
also gave overall negative evaluations to two aspects of current GNSO Website
provision, as shown in Figure IS below, those relating to engaging extemal groups and
building up the profile of GNSO. Steps have been taken in the recent months to address
this issue and the current consultation and Iitst signs of design change on the ICANN
Website are a positive development. But there remains much more to do.
Figure 18: How respondents evaluated different aspecs of the GNSO website
Net highest
Rgnked   Ranked |
highest    lowest
score
| relevant re |oxts and minutes available                39 |        31
Keeping up-to-date With relevant GNSO news and            30           1 1               19
issues
Facilitat  ; o|en discussion on ke  issues                       IS          20               -2
Encouraging views and comments from a Wide range
.                                                                                                                  15                3 1                     - 16
of mterests
Bui.l|~ ; the o |anizational |roiile ofthe GNSO            12        36           -24
same Lse ppa Drama gm| Dfcnnsumem members anamamauaxs
3.7 The parts of ICANN's Website concerned With GNSO are currently set up and used
as a Working tool for Councilors and a repositoxy for all policy-related documentation
that is produced. In many Ways they provide avery comprehensive infoxmation
resource that is potentially of considemble value. Openness is also pursued to a high
degree. For instance, Council phone conferences are recorded as MP3 tiles and
deposited on the Website, so that an interested observer could potentiahy replay the
Whole event, should they Wish to do so. There is an extensive mailing list that
impressively has avemged between 1 and 2 postings per day since mid June 2003. All
minutes of Council meetings are freely available. Figure IS above also shows that
respondents to our survey evaluated the internal roles of the GNSO Website more
positively. However, currently accessing and understanding this data is only feasible for
5 1

LSE Public Policy Group                                                                GNSO Review
people With high levels of skills and Who already know What they are doing and What
they are looking for. Many intexviewees Were particularly frank With us about the lack
of the design and poor organization of the Website for outsiders and even expexienced
insiders to cope With. A previous review of the GNSO Council canied out by a
consultant in 2004 recommended that the GNSO Website be overhauled | yet
subsequently no progress has been made. In terms ofbuilding the Wider protile of
GNSO as part of the ICANN organization, the current Web provision is Weak.
3.3 It is generahy commonplace nowadays for mqor commercial, government and other
non-commercial organizations to monitor the use oftheir Website, collect and process
usage statistics or mettics, and then feed these back into their Web stmtegy and ovemll
operational activity. Basic structural and user mettics can provide invaluable
information to an organization about how customers are interacting With them and the
types ofinfoxmation that they tind most valuable. For a policy development body such
as the GNSO, any feedback on Which kinds of policy information people are searching
for Would seem to be an invaluable commodity in terms ofbuilding a policy resource,
attracting further visits to the site, and potentially recruiting new stakeholdexs. We are
therefore sutpxised to find that the GNSO (and indeed ICANN more Widely) do not
routinely collect Website usage statistics on aregular basis, nor use them to inform their
communication and marketing strategies and to help shape their opemtional activity.
3.9 There can be no transparency Without visibility, and for a body Whose centml task is
Intemet govemance that means critically immediate visibility on the Web. Organizations
With imperfect Websites otien react to deficiencies being pointed out by implying that
such problems can be easily comected by a simple application of finances and effoxt.
We do not share this contidence. In our experience Website problems normally reflect
some substantial political problems inside organizations, and they can only be
successfully addressed When such problems are resolved.
Recommendation 7
The GNSO should improve the design and organization oflhe current websile, develop
a websile slmlegyfor wmmwl improvement and growth over the next three years, and
review usage smximcs on a regular bum to check that mgfc to the websile is growing
over time and understand more fully what external audience: are interested in.
52

LSE Public Policy Group                                                                GNSO Review
3.10 We recommend that the GNSO (and necessarily the Wider ICANN) Website is
redesigned, maintaining the Wealth ofinformatiorr currently available at lower tiers of
pages, but also incorpomting a properly designed top three or four levels. The Website
should aim to present systematically GNSO's (and necessarily ICANN's) activities to a
global public in an accessible fashion, but also in some usethl detail. Following on from
recommendation 7 above, Constituency sections should be clearly labeled from the
GNSO homepage along With other relevant sections, such as `Access our policy
reports', `Policy Work in progress', Events and meetings, and so on. Designated ICANN
staff members will need to be assigned responsibility for maintaining the top layers of
pages in a `fit for purpose' state and to ensure that corporate branding is consistent. A
specitic plan and funding for Website improvements need to be incorpomted into
ICANN's opemtiortal plan as a high priority. The GNSO Council needs to review Web
statistics relating to its pages at least armually. It should aim to grow traffic to the site
year on year, and to keep track of (and develop) parts of the site that are most frequently
visited.
Document management
3.11 It is not always easy to establish the narrative ofpoljcy development from
information and documentation available on the GNSO Website. Many policy papers
are labeled in cortfusing or arcane Ways, and the labels are nowhere explained. Many of
our irtterviewees, particularly relatively new participants in the GNSO process, have
suggested that document management pmctices on the GNSO Website currently make it
difticult to get an overview ofpoljcy issues over recent years. Even highly experienced
Council members acknowledged that it takes a considerable amount of time to get
acquainted With the informal protocols and Working procedures of the GNSO. And the
current documentation pathways and numerous iterations ofpoljcy reports otten make it
con.fusi.ng to orientate oneself in terms of overall rtarmtive ofpoljcy development.
3.12 At the start of our research. We trawled the GNSO Website to get a comprehensive
picture of all policy development that has been carried out since December 2002 (see
Annex A, Figure AS for a summary). We found that the GNSO Website provided very
little overall explanation of how documents published on the site related to different
stages of the policy development process. It is otten hard to tell from the headings and
53

LSE Public Policy Group                                                                GNSO Review
introductory text Whether documents are preliminaxy repoxts, final reports, revised final
repoxts and so on. Having spent four months cmwljng over the ICANN and GNSO
Websites for relevant background documentation, We Were able to compile a
comprehensive table of all policy development Work by the GNSO (see Figure 22
below in Part 4). However, it is still not clear to our experienced research team Whether
We have read all relevant iterations of key documents. In cases Where the Website
groups documents under pmticular areas ofpoljcy Work. generahy documents Were
ordered in a chronological order With scant information about Which documents Were
most authoritative. It seems very important that this Work is presented to the Intemet
community in a form Which facilitates getting up-to-speed, establishing What has been
achieved, and What remains to be done.
3.13 In addition some policy naxratives on the GNSO Website seem to get lost. Figure
22 (in Part 4) presents a summary of how long different policies have taken to develop
through to implementation. In some cases, a final GNSO report or the repoxt of a Task
Force Will be published on the Website. But subsequently there Will be vexy little
information about how the Endings or recommendations in this report have been
progressed or What has happened as a consequence. For example, the recent policy
development process Work on the introduction of new g'I'LD registry sexvices Was
formally adopted by the ICANN Board in November 2005, but then there Was a long
gap With no information on the Website about how the implementation program Would
proceed. Only vexy recently has ICANN announced that implementation of this policy
Would take effect from August 2006. The emphasis of the policy process on formal
stages of development make it crucial that newcomexs or interested observexs are able
to get a clear picture at a glance about Where the GNSO (and ICANN) is on progressing
key pieces ofwork.
Recommendaiion 8
Document management within the GNSO need: to be improved and the presentation of
policy development work made much more accessible.
3.14 We recommend that any redesign of the Website include a forensic trawl through
the existing documentation, removal of extraneous documentation from the main pages
(and creation of archives if necessary), and the introduction of cxystal-clear labeling
pmctices. For example, for each policy development process (PDP) that has been
54

LSE Public Policy Group                                                                GNSO Review
carried out in recent years, there should be a page With an up-to-date document
narrative available, documents clearly labeled, and some basic summary information
about the Task Force membership (Where relevant) and the final outcomes that resulted.
We also recommend that any `Access our policy reports' section of the GNSO Website
contain a full list of currently relevant policy reports, coherently titled, dated, available
in HTML and PDF format, and all in one place, ordered chronologicahy.
3.15 Currently there is no one document that can be used effectively to promote or mise
awareness extemally of the type ofpoljcy issues that GNSO handles. This might be the
equivalent of an annual report or a business plan, essentiahy the kind of document that
markets the Whole of an organizatiorfs activities to a general and extemal audience. A
good number ofirtterviewees, Council members included, raised the point that it is
difticult to get potential organizations interested at events and conferences Without
some kind of summary document that encapsulates the Work of the GNSO. There is
currently an Excel table, circulated and updated by GNSO staff giving an overview of
the current Work that the GNSO is involved in. However, as in other cases, this Working
document Would not be at all useful as a marketing tool for an external audience, since
it is not in an appropriate format nor sufticierttly accessible for lay readers. Alter
thorough search and interview discussion, We are unable to tind a document Which
might be used as a marketing and communication tool as much as a thctual description
of the Work in progress and under proposal at the GNSO.
Recommendation 9
The GNSO should develop and publish annually a Policy Development Plunfor the next
two years, to tm both as a slmiegy documemfor current and upcoming policy work
and as a communication: and marketing Ioolfor general consumption outside oflhe
JCANN community. Ix should dovexail wilh ICANN iv budge! and slmiegy document:.
3.16 This change Would set in place a reahstic progmrn ofwork for the GNSO over a
provisional two-year period. It Would help to prioritize upcoming issues, standardize
programs ofwork. and break any mqj or and complicated issues (such as WHOIS
privacy or ]DNs) into modular and digestible churtks ofwork. It should be based on
consultation across Constituencies, but also involving other ICANN supporting
organizations, and perhaps also the Govemmertt Advisory Committee (GAC) and
external national and international policy development expertise. The Work progmrn
55

LSE Public Policy Group                                                                GNSO Review
should be presented as a formal document, breaking down policy development Work
into some key areas, and setting out some broad timefrarnes for completion. It should be
based on notional calculations of an avemge cost per PDP, and tit Within the available
time and resources given the existing GNSO arrangements. The first year of the Plan
Would need to be regarded as firmly committed but the second year could be more
indicative, so as to retain flexibility to respond agilely. It is important to give
stakeholders a clearer picture not just ofwhat GNSO is doing at present but ofwhat
topics it Will cover in the near fhture. Such a Work plan should also help to neutmhze
any controversy about the scope of the GNSO's policy Work. and to ensure commitment
and buy-in across all Constituencies for upcoming Work.
3.17 The GNSO does not currently have a communications or marketing stmtegy. Some
interviewees have suggested to us that given its relatively narrow policy development
role vis-2-vis the ICANN Board, there is perhaps no justification for it to have one.
Under the existing system of GNSO Constituency membership, however, some kind of
communications With Constituency members and indeed potential new members Would
seem to be a core plank of the creating incentives for continued and new forms of
participation. Providing incentives for members to participate (or more fllndzmentahy
to pay their dues) is certainly a central plank of almost all membership-based
organizations. As illustrated in Part 2, Constituency membership lists and
communications With members are largely disaggregated across Constituencies, and
consequently there is no centml register of GNSO members and certainly no proactive
Work to keep members up-to-date With GNSO activities and issues. Neither the GNSO
nor ICANN appear to use RSS feeds With stakeholders. As a result, the onus is on
stakeholders to keep up With events through trawljng the GNSO Website.
3.13 Given that GNSO councilors and Constituency activists voluntarily commit
thousands of hours of their time to policy development Work. it seems strange that there
is currently no policy publication series or any sign of ICANN branded output that is
linked to GNSO or represents its activities to any Wider public. For example, in a
compamtor organization the e-Business, IT and Telecoms (EBITT) Commission Within
the ICC produces relatively short PDF policy reports or digests, consistently bmnded
With an ICC header but presented as the Work of the EBITT commission. GNSO policy
reports tend to come in all shapes and sizes, Without consistent ICANN branding. Our
56

LSE Public Policy Group                                                                GNSO Review
previous recommendation in paragmph 2.22 relates to greater use of `off-the-shelf`
tools and coxpomte matexial by Constituencies in order to strengthen the Working and
marketing ]j.n.k between the GNSO and ICANN as awhole. Establishing ICANN policy
or Working paper seiies might be a fmitful opportunity to bland existing Work by
Constituencies and other committees in a more coherent and accessible Way, and to
develop more gradated outputs from the policy development process. Gmdated outputs
might Well focus on different potential audiences for policy Work. mnging from formal
PDP output, through to press and media Iiiendly output, and shoxt and digestible thj.n.k-
pieces or discussion pieces that could be circulated to universities, non-govemmental
and research sectors, and other Intemet stakeholders.
Recommendation 10
The GNSO and JCANN should work proaclfvely to provide information-based
incemfvesfor stake}/[older organization: to monitor Mdpmicwwe in GNSO issue:.
3.19 In Pan 2 We recommended a fundamental change to the Way in Which
organizations intexface With the GNSO and ICANN, by shifting from a GNSO
Constituency membexship model to an ICANN stakeholder model. An ICANN-level
interthce in our view Would greatly enhance the prospects of involvement by a Wider
mnge of organizations in GNSO's policy development, and it should create new
possibilities for growing clearer incentives for new organizations to take part as ICANN
stakeholders. There Would be obvious opportunities here to develop a comprehensive
database or register of ICANN stakeholdexs, to segment ICANN stakeholders according
to their interests or extent ofpaxticipation, and then to develop `selective incentives' for
these organizations in the form of regular and accessible infonnation, news alerts,
policy digests and so on. An ICANN stakeholder register Would promise a mnge of
Wider marketing benefits, Would allow for more coherent communication across
different ICANN areas. There is avast Wealth of institutional and technical knowledge
Within the ICANN community, and amore integmted communications strategy at
ICANN level could help to feed a much greater mnge of this knowledge into GNSO's
policy development Work. An ICANN Working paper sexies on gTLD issues Would be
one example of a selective incentive for organizations to become ICANN stakeholdexs,
and get more deeply involved in policy development Work. In the contemporaly theory
of interest group and NGO paxticipation, making available to membexs `selective
57

LSE Public Policy Group                                                                GNSO Review
incentives' that are not available to non-membexs is assigned a great deal of sahence.
GNSO and its Constituencies need to enhance their efforts in this respect.
The visibility ofthe GNSO within the ICANN community
3.20 In addition to the problems clouding GNSO's external visibility, there are also
impoxtant issues about how visible or We]1-undexstood its Work is even Within the
confines of the ICANN community. There Were indications that GNSO in some
respects has alow profile role even Within the ICANN community itself For instance,
our interviewees' perceptions tended to vary Widely on how high protile the GNSO is at
ICANN Board level. We asked interviewees to estimate roughly What percentage of
Board time is taken up dealing speciiicahy With policies developed by the GNSO.
Responses vaxied from around 30 per cent to around 10 per cent. Some people have said
to us that the GNSO is a vital influence in the BoaId's deljbemtions on gTLD issues,
and indeed our suxvey respondents scored `Delivexing practicable recommendations to
the ICANN Board' highest in a list of aspects of the PDP. Other intexviewees have been
uncertain about how important the GNSO has been on key issues, notably IDNs and
policies on selection ofnew g'I'LDs.
3.21 In order to create a clearer picture of how much Board time is spent discussing
policy recommendations speciiicahy originating from the GNSO, We did some analysis
of a]1 available Board tmnscxipts and minutes since 2002. We counted the total number
ofwords in each document, and estimated What proportion ofwords related to specific
GNSO policies, and What propoxtion ofwords related to Wider gTLD issues (not
necessmjly emanating from the GNSO). Figure 19 below suggests that around one third
of Board discussion relates to g'I'LD issues. The estimated proportion of GNSO-specific
discussion however is significantly less; never more than S per cent and 2 per cent on
avemge for 2003-04 and 2004-05 (for more on both these categoxies year-on-year,
please see Annex A. Figures A3S and A39). This seems significantly less than the 10
per cent lower estimates that We Were receiving from more skeptical intexviewees. We
cross-checked these percentages With the percentage ofwords in the ICANN adopted
budget relating speciticahy to the GNSO. We found here that there has been a gradual
increase to 10 per cent in 2006-07, providing some rough corroboration for the visibility
tbctor of 10 per cent for the GNSO at Board level.
53

LSE Public Policy Group                                                                GNSO Review
Figure 19: Estimating the importance of GNSO issues in ICANN Board level
meetin |s
Rgws Show % Oftm                                 2002-    2003-    2004-    2005-    2006-
03                04                0 5                06               07
Estimated proportion of ICANN Board
meeting discussion relating to gTLDs             45          23          35          36
Estimated proportion of ICANN Board
meeting discussion relating to the GNSO                    2           2
Estimated proportion of ICANN annual
adopted budget documents relating                  3            4            2                         10
|eciticall  to the GNSO
NOTE: Rows 1 and Z in this table show results ofbasic textual analysis of transcripts and minutes of
ICANN Board meetings (see Figures A38 and A39 in Annex A for more details). Row 3 shows results
from a similar analysis on Adopted Budget documents available on the ICANN website. We reviewed
these documents for direct references to the GNSO and the gTLD supporting organization.
same Lse ppa anal as DHCANN said mean| uansm|ts and mmutes
3.22 Achieving balanced growth in the relationship between the GNSO and ICANN as
awhole Would seem to a sensible objective to aim for in terms of development over the
next ten yeals. It is interesting to look at data for time spent by the ICANN Board
discussing issues/policy Work emerging directly from the GNSO, and compare these
percentages With total ICANN expenditure on GNSO as apercentage of total ICANN
expenditure (see Figure S above in Part 2). We might expect to see abroad balance
between these two percentages, suggesting that the ICANN Board is roughly `getting its
moneys Worth' in terms of the amount of expenditure it is allocating to the GNSO and
the policy development returns it is seeing on this investment. In terms of balanced
development, these figures seem quite encoumging. There currently seems to be a 'IO
per cent trend', the consequence of gradual growth over recent yeaxs. We estimate that:
-   ICANN expenditure on the GNSO is currently around 10 per
cent of total expenditure;
-    ICANN spent roughly S per cent of its time in 2005-06
discussing policy Work of the GNSO;
-    around 10 per cent of ICANN budget documentation relates
speciIica].ly or indirectly to the GNSO and its activities.
It Would be important to keep under review indicatoxs Which give an idea ofretum on
expenditure in this Way | a kind of `bangs per buck' guide to balanced growth. Our
59

LSE Public Policy Group                                                                GNSO Review
indicators are relatively broad estimates, and there is obviously room for refinement of
such indicators.
3.23 Tmnsparency in the relationship between the ICANN Board and the GNSO Was an
impoxtant concem in many interviewees' comments to us. Decision-making processes
Were alien seen as rather opaque, and channels of communication between the Board
and the GNSO Were seen as lacking. The GNSO Council cumently elects two Dixectors
to the ICANN Board by mqjoxity vote. Although these Dixectors have played the role of
liaison between the GNSO Council and the Board in recent years, they do not have a
formal role as representatives of the GNSO duting Board delibemtions. All ICANN
Board membexs have a duty to act in the best interests of the organization as awhole
and its mission, rather than to represent particular interests. At times the
communications and undexstanding between the Council and the Board have been
stmined. Recent controvexsy over the Way in Which the Vexisign settlement Was handled
mises serious questions about the quahty of communication between the ICANN Board
and the GNSO. Many views expressed by Council membexs duting our intexviews, and
indeed much of the Wxitten documentation on this paxticular decision, suggest aneed for
more foxmahzed and transparent, two-Way channels of communication on how GNSO
views are taken into account and how decisions at Board level are made. Figure 20
below shows the number of GNSO Council meetings at Which Board members have
been present since 2002. The number of meetings attended per year has increased from
zero out of 13 meetings in 2003-04 to eight out of1S meetings in 2005-06.
Figure 20: Attendance of ICANN Board members at GNSO Council meetings
2002-     2003-     2004-     2005-     2006-
03                04                0 5                06               07
Number of Council meet |s              IS      13      13      IS      4
Ax which ax lean one Board member was                                      4                         2
' resent
Ax which ax lean two Board member:                                                     2
were |resent
NOTE: Figures in this table cover January ZOOZ to May Z006. The number of Council meetings in ZOOZ
refers specifically to the number of meetings of the Names Council ofthe foxmer DNSO. Meetings
include face-to-face and teleconference call s. The number of Council meetings listed is based on
meetings for which data was available.
same LSE ppa and as Dfamso cnumx mean| attendance
60

LSE Public Policy Group                                                                GNSO Review
3.24 Yet there are very few foxmahzed and institutional channels through Which the
GNSO Council may communicate With the Board and other senior representatives
across ICANN organizations. Our interviews suggested that the cument
communications channels between the Board and the GNSO rely a great deal on the
constructive Working relationships established between the cument Chair of the GNSO
Council, the ICANN President and Chief Executive, and key Board membexs. The
GNSO Chair is a very important and intemally influential figure Within ICANN, Whose
agenda management capabilities play a key role in progressing GNSO's Work and
tbcilitating constructive debate. The Chair does not cutrently attend Board meetings,
and there is no automatic progression from the position of GNSO Council Chair to a
place on the Board. This lack of foxmahzed communication channels (and the
consequent emphasis on informal channels between the GNSO Council Chair, the
ICANN President and Chief Executive, and the Board) may be a relatively high Iisk
stmtegy to maintain for two reasons. First, the system is reljant on constructive Working
relationshjps between senior pexsonahties. Although the current Council Chair has
ready access to the ICANN Chief Executive, this axrangement is rather infonnal and
might not continue With other pexsormel in these posts. If personal relationships break
down or do not fimction adequately, there are few institutional safeguards to protect the
interests of the GNSO and its Council as a Whole. Second, the lack of open and
regulaxized communication channels between the GNSO and the Board increases the
risk of perhaps urmecessaly resentment and confrontation at GNSO level, such as
sutfaced in the atiexmath of the 2006 decision on the Vexisign settlement.
3.25 Sub-units Within complex organizations are genemlly made more visible and
salient When they have a clear personal embodiment, that is, someone Who clearly
represents them in other important fomms. Some interviewees strongly suggested to us
that the role of the Chair of the GNSO Council should be enhanced so as to help GNSO
become more influential. For instance, the Chair could act as a non-voting liaison to the
Board, or he/she could automatically take a seat on the Board for two yeaxs alter their
Council term had ended. Another potentially useful idea put to us is that an overall
ICANN central executive committee should be established, perhaps including the Chair
of the Board, the President and Chief Executive, the GNSO Chair and chairs of other
Supporting Organizations, and the chair of the Government Advisory Committee. This
central executive committee Would be aregulaxized and minuted discussion ofwork
6 1

LSE Public Policy Group                                                           GNSO Review
going on across ICANN and GNSO and other represented bodies, and give a chance to
maintain open channels of communication on the many cross-cutting issues now in
discussion.
Recommendation 1]
The position oflhe GNSO Council Chair need: to become much more visible within
JCANN and to Carry more institutional weight.
3.26 It is important to stress that this is an area Where We cannot put forward very
specitic recommendations, because ICANN's central organization Was outside the
scope of this Review. In addition, the role of Council Chair is already onerous enough,
so that adding any new duties or roles Would require a slimming down of other
activities (such as Council teleconferences, see below). But What is not sensible is to
continue With the GNSO Council opemting as alarge body With only a diffuse identity
elsewhere in ICANN and With an apparently `submerged' or under-visible leadership.
3.27 Tuming to GNSO's internal rules, a transparent organization is also one Where the
reasons for people saying What they do are Silly undexstandable to internal and external
obsexvexs. In contexts ]j.ke GNSO, Where decisions can otten impact on the economic
life chances and tinancial situations ofparticipants in decision-making, it is particularly
impoxtant that people acting in representative roles should meticulously document and
declare any relevant interests that they may have. Somewhat under half our
interviewees complained strongly to us that in their view disclosure of interests by
GNSO Councilors Was not always upheld. They felt that it Was otten impossible to tell
Which organizations individuals Were representing. Other intexviewees argued that the
cument approach ofrelying on self-regulation of disclosure Works relatively Well,
paxticularly given the close knit community of ICANN, the few degrees of sepamtion
between most pxincipal actors, and the intensity of discussion between them. The
general view amongst councilors themselves is that any compromising relationshjps and
interests Will be uncovered one Way or another. Yet however useiijl these self-
comecting tendencies may be, suspicious of confhcts of interest are currently Widely
held elsewhere in the ICANN community. Especially in tandem With perceived
cliqueness in some constituencies, these perceptions are highly corrosive of the
legitimacy of GNSO's delibemtions.
62

LSE Public Policy Group                                                           GNSO Review
Recommendaiion 12
The policies on GNSO Cozmcilor: declaring interests should be strengthened. Provision
for avole of (no confdencey leading to resignaiion should be imroducedfor mm-
compliance.
3.23 We recommend strongly that the GNSO Council establishes awxitten policy on
disclosure of interests, circulated to each new member and to existing membexs
annually. The Council should discuss and agree some mechanism of enforcement
requiting a vote-of-no-conjdence in the Council With a quahiied mqjoxity (15 or more
membexs) if interests are not declared. This extm incentive Would help encoumge
membexs to comply.
3.29 Where the number of active membexs in Constituencies become low or vexy low,
there is a risk that those Who sustain the Constituencies participation in GNSO
processes will be perceived extemally (lightly or Wrongly) as running a clique-
dominated process, one perhaps unreceptive to the involvement of new actoxs or
players. The current GNSO Bylaws stipulate that any Councilor serves for atwo year
term, With a staggered changeover of two of a Constituency's Councilors in one year
and then one in the next, and so on. Some Constituencies, such as the Registmrs,
stipulate term limits for Council representatives. However, councilors in other
Constituencies may currently continue to serve for more than two years (indeed they
may sexve indefinitely) if re-elected by a mqjoxity oftheir Constituency members. Our
analysis of tumover ofindividuzl GNSO Councilors terms of service since 2002
63

LSE Public Policy Group                                                                GNSO Review
Figure 21: Turnover ofindjvidual DNSO/GNSO Councilors, since January 2001
Number of individual        Number of             Number of
Council membexs          changes of           changes of
since Jan 2001              representatives          representatives
since Jan 2001     since Jan 2003
NCU           IS         5 |
|TLD | |    2 |
|     11         3         5
Re | strar                                                 7                                               2                                            2
BC                    5               2              1
ISP             5         2 |
NOTE I The frgures rr. thrs tame covertheperrod from january zom to june more column I shows the total
number Of drffererrt rrrdwrduats Who have Sewed as a Member Of erther the Names courter! (error to end zooz)
Or the GNSO courter! (srrrce begrrrrrrrrg zoo;) colurrrrrs z and 3 Show the number Of changes Ofpersorrrrel that
took place rr. the Names courter! up to jar. zoos, and then subsequerruy rr. the GNSO courter! Srrrce jar. zoo;
NOTE z The Ip corrstrmerrcy porrrted out merr retatwety hrgh raw oftumover Ofcourrcrlors rs due to the fact
that aaerthe decrsrorr to more from a systan Of 3 courrcrtm to z courrcrlors per corrstrmerrcy, rt took the
decrsrorr to rerur. electrorrs for rts represerrtatwes to the courter!
Source LSE pm anal Srs ofdata Or. terms Ofsmrce OEGNSO courter! re|reserrtatwes
shows awide vaxiation across Constituencies (see Figure 21 above). In the case of the
Business and ISP Constituencies there has been vexy low tumover ofpexsormel amongst
their representative on the Council, and only five people have sewed as one of their
councilors since 2001. At the other end of the spectrum, some Constituencies show
compamtively high rates oftumover, such as the Non-Commercial Users. Some
interviewees have suggested that this is due difliculty in retaining Counciloxs long
enough and the fact that NCUC Councilors are alien not Iimded. Other NCU
representatives have suggested that high tumover reflects an open and dynamic culture
Within the Constituency. In addition to achieving some reasonable degree of cixculation
amongst GNSO counciloxs, it Will also be important to have sufticient people involved
in how Constituencies are run to prevent their oflicers being the same people for long
periods of time. In addition to achieving some reasonable degree of cixculation amongst
GNSO councilors, it is also important to have sudicient people involved in the running
of Constituencies so as to ensure stability When people change jobs and therefore leave
the constituency, etc. Thus areasonable pool of involved leadexs is an impoxtant goal
for the long texm health and fimctionjng of any constituency.
64

LSE Public Policy Group                                                                GNSO Review
Recommendation 13
Fixed term limits should be imroducedfor GNSO Cozmcilor: either ofrwo two-year
term: (as applied in some Constituencies already) orperhaps ofa single three-year
term.
3.30 Because of the small numbers of councilors in some constituencies and the
potential for delegitirnizing perceptions to axise, there should be atenn 1i.mit ofthiee or
four years, alter Which councilors must step down from membexship for at least aterm.
Standardizing atenn limit of two two-year terms Would be one solution. Individuals
Who reach the end of their terms Would need to step down from the Council until anew
election Was held. Given the regional representation that Constituencies aim for this
Would in practice be two yeals off (so a four years on/ two yeals off cycle), unless the
pexson changed region. Which does happen. Some interviewees suggested to us that
councilors should sexve only a single three year term and then step down for a year, an
axrangement that With staggered retirements Would also allow for more stability in
Constituency representation. One possible disadvantage Would be that With regional
representation a leaving councilor Would not in practice be able to retum for three yeals,
When their region's seat Was once again being contested, creating a 3 yeals on/3 yeals
off cycle, again unless the pexson changed region. Both GNSO Council and the Board
Will need to consider Which of these options to pursue, having regard to the need to
marshal and conserve expertise While also avoiding any creation of cliques. It should be
bom in mind, however, that individuals Who reach the ends of their term could still be
active in the leadership of their Constituency and in task forces, so that there is no need
for their talents or commitment to be completely lost to ICANN in an `off peiiod.
65

LSE Public Policy Group                                                           GNSO Review
Part 4:
How effective the GNSO has been in undertaking its
work and developing policy positions
The operations of the Policy Development Process
4.1 The GNSO Council discusses issues associated With genexic domain names that are
relatively complex and alien quite technical. The GNSO Council has a set of formal
procedures for looking at new issues, seeking views from the Constituencies and other
stakeholders in the Intemet community Who Want to comment, debating the views
expressed in the Council, seeking common ground and ttying to reach a consensus view
of the issue. These procedures are called aPo]icy Development Process (usually
referred to by the initials PDP Within ICANN). The stages and timings for a PDP are set
out in the ICANN Bylaws. As Figure 22 below sets out, the GNSO has been involved in
around 12 mqjor pieces ofpoljcy development Work since December 2002. This policy
development Work has dealt With potentially highly technical issues, Which When
reviewed for Wider relevance to society at large, reveal some direct implications (many
at an evexyday level) for Intemet usexs and sexvice providers. We bxiefly summaxize
these in the next few paragraphs.
4.2 Policy Work relating to the WHOIS database is apxime example of how such
naxratives could be made more accessible and transparent to interested newcomexs or
obsexvexs. Since December 2002 the GNSO has conducted 12 separate policy
development processes ranging across four geneml areas (see Figure 22 below). Policy
Work relating to the procedures and timctions of the WHOIS information sexvice has
accounted for seven out of the PDPs ljsted. Task Forces have been established,
combined, disbanded, and even alter close analysis, it is diflicult for our research team
to establish exactly Where issues currently ]je, and What might be arealjstic ttmeframe
for fixture progress. WHOIS has involved ahuge amount oftime and expertise from
GNSO stakeholders. We estimate very roughly that this policy development Work on
WHOIS has involved roughly 39,000 hours of Task Force member and GNSO
Councilor time since the tirst Task Force Was launched in 2001.
66

LSE Public Policy Group                                                           GNSO Review
Figure 22: Summary of the GNSO policy development work since December 2002
GNSQFOJW Wmk                        PDP              Task Force    Adopted          Policy
launched        or final         by Board    Implemented
re |on
WHOIS Review                    __ May os       Nov 04
Inter-re|istmr Transfer Poljc    _ Feb 03         A|  03      Nov 04
Expired Domain Deletion         Sept 02         June 03         Sept 04      Nov 04
Polic
WHOIS Task Force 1              Oct 03
WHOIS Task Farce 2               Oct 03                   Tm Form Combined we Mow;
WHOIS Task Force 3             Oct 03
New Registxy Sexvices              Nov 03          Jun 05          Nov 05       Plurmedfor
2006
GNSO resolution on new .net    NA               Aug 04
a|erator
WHOIS Combined 1             Dec 04         A| 05         Tmk Forces Cumbznm we
WHOIS Combined 2              Dec 04          Se |t 05                    MOW)
Purpose ofWHOIS and           Jun 05          Mar 06
WHOIS contact
Introduction Ofnew |TLDs       Dec 05         ___
Amendments to conditions for  Feb 06
existing gTLD Registry
contracts
same Dmmmtaunn an n1excANNWebs=te
Applying anotional `cost per hour' figure ofUSD ISO, the total policy resource
expended on WHOIS policy since 2001 amounts to just under USD 7 million (see
Annex A, Figures AS and A6 for details on how these estimates Were calculated).
4.3 The GNSO has also developed procedures to safeguard the integrity of domain
names, paxticularly the Way in Which market interactions between registration
stakeholders impact on the domain name holdexs. The policy on inter-registmr transfexs
developed standardized procedures covexing the Way in Which aregistered domain
name may be tmnsfemed from one registmr to another. This policy established
procedures for allocating administmtive obligations on `gaining' and`1osing' registrars,
so as to reduce duplication ofworkload and hannfijl consequences to the registmnt. The
policy on `expired domain name deletions' sought to provide safeguards for registrants
in cases Where a domain name registmtion pexiod expires, and Without the oxiginal
registmnfs knowledge the domain name is put back on the market and purchased by
67

LSE Public Policy Group                                                                GNSO Review
other registmnts. This policy Work set some ground rules for notification of the
registmnt and aredemption grace period.
4.4 In 2004 the GNSO initiated policy Work on procedures to ensure that gTLD
Registties notify ICANN of any new services or changes to sexvice delivery that are
introduced. The final report Was published in June 2005, setting out consensus policy on
the steps that should be followed by gTLD Registxies and ICANN staff for reviewing
and agreeing to specitic changes to existing sexvices or introduction of new services.
More recently, in February 2006, a much larger and potentially more controversial
piece ofpoljcy development Work has begun on possible improvements that could be
made to conditions specified in existing gTLD Registxy contracts.
4.5 In December 2005 the GNSO initiated new policy development on procedures
governing the introduction of new genexic TLDs. This new PDP examines What types
ofgenexic TLD are most desimble, how many should be introduced and duting What
time span, and by Which allocation mechanism new TLDs should be introduced. This is
cumently one of mqor pieces ofpoljcy development under Way in the GNSO. Related
to new gTLDs, but not quite under Way as formal policy development Work, is the issue
of Intemationahzed Domain Names (again otten abbreviated to ]DNs). The Iitst
ICANN committee Was formed back in September 2001 to discuss all issues relating to
the introduction of IDNs. The ICANN President's Advisoxy Committee announced at
the staxt of 2006 that ICANN Would be running a test-bed for the Iitst IDN prototype
forms. Now the GNSO Council is currently in the process of setting up an informal
Working group to consider key policy issues that Will axise from this Work on ]DNs.
4.6 The use of open access mailing lists is arather distinctive feature ofthe Way in
Which the GNSO operates. Lists provide a vital Working tool for GNSO stakeholders,
not to mention a vast source of information for potential paxticipants and observexs Who
have enough patience and expextise to follow policy development discussions. Open
mailing lists give the impression of a highly tmnsparent organization, despite the caveat
that open mailing lists can be used for strategic purposes and alarge amount of
comespondence takes place off list. A review of the GNSO Council's main public
mailing list shows that it has averaged (mther impressively) at least 2 or 3 postings per
63

LSE Public Policy Group                                                                GNSO Review
Figure 23: Main themes discussed on the GNSO Council mailing list, by year
General Council bustness
Other
ICANN Wade /Board /DoC (mclude budget)
WHOIS related
Relattng to ICANN In gmaal
GNSO review / GNSO opemttonalplan/PDP related
New Regxstry services / Exxstmg Regxstry contracts
Introduction of new gTLDs
All Counctl votmg
Intemattonahzled Domain Names (IDN)
dot net dot cam, Sxteftnder
WIP O / UDRP
GAC, ALAC, ccNSO related
0              5             10            1 5            Z0            Z5            30
Percentage ofposts to the GNSO Counctl mathng list smce July 2003
D Z003        D Z004        I Z005        I Z006
Suurce LSE FFG analysts ufthe GNSO pubh: manhng hst June ZEID3 tu June ZEID6
day since June 2003 With at least one third of total posttngs dealing With geneml or day-
to-day business. In terms of substantive policy development, Figure 23 shows that Work
relating to WHOIS ranks highly in terms of coverage, accounting for just over 10 per
cent of all postings to the list since June 2003. Other mqjor policy development issues
have been Work relating to the introduction of new registty services and more recently
procedures for managing the introduction of new genetic Top Level Domains.
4.7 We noted above the importance of the formal Policy Development Process (PDP),
the set of formal procedures used by GNSO Council for looking at new issues. The PDP
rules govem how the Council requests views from the Constituencies and other
stakeholders in the Intemet community Who Want to comment, debates the views
expressed, and seeks common ground in an effoxt to reach a consensus view of the
issue. The ICANN Bylaws lay down demanding deadlines for how long aPDP process
can last, designed to help ICANN respond at `lntemet speed' to problems or issues and
69

LSE Public Policy Group                                                                GNSO Review
to avoid the long timeljnes sometimes perceived as charactexistic of governmentally-mn
international fomms and bodies (see Annex A. Figure A46 for a stage overview of a
PDP). We encountered arange of views about how Well the PDP is Working. Most
cument Council membexs view the process as genemlly effective. However, many other
interviewees Were cxitical or even scathing about how PDPs operate, With some
expexienced ICANN obsexvexs describing it as `Byzantine' or `archaic'. For outsiders, it
is a complicated process to understand and intemahze, punctured by sttict time
deadljnes for paxticular stages, regular voting, and public comments pexiods. Our
analysis below shows that PDPs valy greatly in the quality and diversity of comments
received.
4.8 We asked our survey respondents to mte different aspects of the PDP in terms of
strengths and Weaknesses. Figure 24 below identities the highest and lowest scoxing
aspects. Respondents Were generahy positive about the extent to Which the PDP
delivexs practicable recommendations to the ICANN Board, and the Way in Which the
GNSO makes the best use ofpoljcy suppoxt functions, perhaps reflecting the recent
strengthening of staff support Iimctions to assist GNSO in the last year or so. Three
mqjor Weaknesses Were raised by insiders. It is otien difticult to keep to the time
schedules outlined in the ICANN Bylaws. There are diffexing views of how best to
dmw on and incoxpomte extemal expertise in the Council's policy development Work.
And as mentioned in Part 2, difliculties have been encountered under cument
axrangements in ensuring that Constituencies provide adequate documentation on the
number of members Who have participated in detinjng a position they have adopted.
4.9 The global context of the GNSO policy development Work and the need to
coordinate input across awide mnge ofdifferentstal<eho1ders increases the impoxtance
of relatively formahzed management of time schedules and delivery. The PDP is
nothing if not prescxiptive about time periods allocated to different processes. As one
interviewee neatly summed up: `the PDP is about delivery by management of process
mther than management of ouwul:'. Despite the current stress on aprescxiptive
approach to process management, We found a geneml acceptance that the PDP does
struggle to stick to time schedules set out in the Bylaws. According to the Bylaws the
70

LSE Public Policy Group                                                           GNSO Review
Figure 24: Survey respondens' views on different aspects of the PDP
Ranked        Ranked         Net
hi|hest         lowest         score
Delivexing practicable recommendations to the
ICANN Board                                                                    24                   1 1                 13
| the best use of |o1ic  su|oxt functions          22             12            10
Pic ~ ; the Ii| tissues                              20          16         4
Sco|~  ; |oljc  work .v= a o| ~ate1                              14               16             -2
Identi   ~  ; issues earl  eno  ;                                  15               20             -5
Ens-ming that the PDP incorporates the Widest
a . ticable I| | ofviews                           15          21
| use ofextemal e  |extise and research            20             30           -10
Stickin| to . ; eed time schedules                              11               25             -14
same Lse ppo Drama gm| Dfcnnsumem members anamamauaxs
PDP is supposed to last around 150 days from the initial request to a Board adoption. A
quick glance at the ovexview of policy development Work in Part 1 of this repoxt
suggests that in the few cases Where this is measumble, the PDP has taken up to 2 years
or more.
4.10 In most organizations there is inevitably some degree of divergence between
theory (or regulation) and practice. Councilors suggested to us that the GNSO Council
is We].1 aware that it does not stick to time schedules specified in the Bylaws, but it does
tend to stick to more informal and specitic schedules that are determined consensuzlly
in the Council. These `effective' timings respond to the diffexing levels of difticulty
associated With individual PDP topics. A limited review of the GNSO Council by a
consultant in 2004 recommended that the stages of the PDP should be revised and
claxitied. Since then there are clear signs that the GNSO has made some progress to
address this issue. Nevertheless, under the current armngements there are some good
reasons to think that newcomers or potential paxticipants in the policy development
process might We].1 be confused or put offby the Wide divergence between What is
formally supposed to happen and What happens in pmctice on the timings of stages. For
prospective membexs Wishing to get involved in policy development, just getting
acquainted With the different arrangements framing the PDP involves a steep learning
cutve. Having intemahzed the rules involved, potential participants are then fhced With
the prospect that these rules are genemlly viewed as largely symbolic, and that in
general, the PDP can take up to 2 yeals in reality. As one representative from aFortune
500 corporation told us: `it may not come as a surpxise that commercial corporations
71

LSE Public Policy Group                                                                GNSO Review
choose not to get involved in aprocess Which can take up to two yeaxs for any results to
Biter through'. There is also a risk With such large divergences that GNSO becomes
habituzted to crashing foxmal deadljnes, and thereaher treats all deadlines as elastic. We
looked carefully for but did not detect such an attitude, but instead a general concem to
expedite mattexs as fast as feadble. None the less, in general it is desirable for more
effective and reahstic timings for PDP phases to be formulated and then Wherever
possible stuck to.
4.11 In terms of staff support, there has long been provision for secretaxial, recording
and agenda support to the GNSO Council. As its revenues have grown in recent yeals
ICANN has considerably increased the numbers of its policy suppoxt staff centrally. In
the last year (2005-06) much more staff suppoxt has been allocated to the GNSO
Council supplementing previous provision. Our analysis of attendance at the GNSO
Council meetings shows clearly the increasing level of attendance of ICANN staff since
early to mid 2003 (see Figure 25 below). ICANN policy staff now play an enhanced
role in informing and assisting the Work of the Council. There is already evidence from
our interviews With Council members and staffthat a more intensive resourcing and
management of the PDP has had a positive impact on pexformance. A Wide range of
Councilors mentioned identitiable benefits of more intensive staff support, paxticularly
more systematic approach to gathexing and processing Constituency statements,
keeping policy development on track, tbcilitating dissemination of materials, and
shaping Iinal repoxts. Greater staff suppoxt should thciljtate quicker time-to-market and
delivexy of outputs potentially Within a year, Which may help to bring actual PDP
timeframes closer to those set out in the Bylaws.
4.12 One of the strongest themes to emerge from our intexviews on the PDP relates to
the distinction (and ]j.nk) between management of delivery by process and management
of delivery by outputs. The language of the Policy Development Process is
programmed from the start to emphasize process rather than outputs being achieved. In
the last year ICANN staff has Worked intensively to grow the use of project
management methodologies across the organization, and there is clearly great potential
for the more systematic application ofproj ect-based Working to the policy development
Work of the GNSO, that is, With a much more explicit link between a managed process
72

LSE Public Policy Group                                                           GNSO Review
Figure 25: Attendance of ICANN staff and other ICANN representatives at GNSO
Council meetings
O     as
O |
|     an
 |
|     15
 |
|
|
|      5
Z      [I
la    zna    In    zna    In    zna    la    zna    In    zna    In
Hatfyea pennas (swung tan and M)
a attso cnunttmembm               I ICANN Ms
a other tcAM mgantsatnns            I ICANN BUM members
a other Relevant
same LSE ppa analysts Dfattso cnumt meeung attendance
and a set of outputs. It might be Woxth changing the meaning of the PDP to denote a
`Policy Development Project X, With the emphasis upon achieving agreed change, mther
than PDP denoting apmcess.
4.13 A large mjnoxity ofinterviewees experienced With the PDP argued that the issue
repoxts in the past have tended to summaxize different points of views rather than to
identify and provide a preljminaxy discussion of the core issues and potential divisions
that need to be bxidged. In their view, issue reports should be premised on and Wxitten
around core issues, mther than focusing on presenting the divergent mnge of opinion
across relevant stakeholders (not just Constituencies). The suggestion is that the issue
repoxts should be dtiven by getting to gtips With the pressing substantive issues as
outputs, rather than (as now) going through the process of collecting views from all
Constituencies and then presenting them to the Council. As one expexienced
commentator put it: `it is about asking the light questions early enough'. The recent
PDP on the putpose of WHOIS illustrates how potentially intractable divisions can
pexsist if the core policy issues are not isolated early on in the process, and allowed to
73

LSE Public Policy Group                                                                GNSO Review
dtive the consequent discussion. Most project management methodologies strongly
emphadze issue analysis as a dtiver for data collection and research, rather than issues
being the product of data collection and research.
Recommendation 14
The GNSO Council and relaxed policy sldshould work more closely together to grow
the use ofproject-management methodologies in policy development work particularly
focusing on how targeted issue analysis can drfve dam colleclionjhzm smkeholders
(rather than vice versa).
4.14 The application of project management methodologies to the Work of the GNSO
Council and its Constituencies could greatly enhance the effectiveness and the speed of
policy development. We Would recommend that project management training is rolled
out across Council membexs and relevant policy staff. Advice taken on this issue
suggests the costs of such training Would not be unreasonable. Project management
methodologies Would help to de-emphasize adherence to process and achieve a more
effective balance between envisaging outputs and managing required processes. Issue
repoxts produced by policy staff should focus on targeting the `crux' or core issues or
potential divisions at stake, and present parametexs for potential negotiation and
agreement between stakeholdexs. Cutrently issue reports tend to simply descxibe the
mnge of existent views on a subject, and tend not to synthesize views down to issues.
This enhanced analytical requirement Would require a high level of trust between the
GNSO Council and policy staff as policy staff Would be framing issues more explicitly.
Some signs of tension and lack of communication between policy staff and the GN SO
Council Would need to be addressed as pan of this growth.
4.15 An underlying problem in developing better Working may be that in quite a range
of our intexviews With GNSO insiders, We identiiied some unusual tensions in the
relationship between the staff assisting GNSO and some Council members. There has
apparently been a strong feeling in parts of the Council that the policy staff owe their
pximary allegiance to ICANN management, Who appoint them, rather than to the
Council itself. Some counciloxs have a strong preference for a do-it-youtself approach
to policy-making, believing that almost all stages of the process should be undextaken
collectively by Council members drawing on their own expertise and contacts With
Constituencies to determine What the issues and evidence are, What confhcts exist and
74

LSE Public Policy Group                                                                GNSO Review
how they should be resolved. This approach is resistant to too much ICANN staff
intervention as undermining GNSO's independence. It is distmstful of outside expextise
as injmical to bottom-up stakeholder involvement, a stance also reflected in a tendency
for task forces to be staffed by GNSO counciloxs or active Constituency membexs.
Some interviewees argued that a formal service level agreement between the Council
and ICANN staff is needed, Which Would tightly specify What Was to be done and in
What timefmmes. A mnge ofintexviewees suggested to us that because of the histoxy of
these past attitudes Council members can alien take staff for gmnted and urmecessaxjly
limit their role in informing discussion. By contrast, a mqjoxity of other Councilors
Wholeheartedly supported a strengthening of the policy expextise available to the
Council and believe that too little staff time has been allocatedto the needs of the
GNSO Council vis-2-vis Wider ICANN functions. In extensive interviews With staff We
did not find a conflict of loyalties between an ICANN or a GNSO orientation, but noted
instead a strong and neutral dedication to helping GNSO Work in a more expert,
effective and speedy manner. We do not thj.n.k that a framework agreement between the
Council and staff is necessary or helpful: creating it Would sexve more as the symptom
of a problem mther than the cure. However, there does need to be a review and
claxitication of the basis on Which staff attending GNSO meetings operate, especially
for those Working extensively or pximaxjly for GNSO. It is impoxtant to provide some
additional assurance to GNSO stakeholdexs that staff Working to support both the
Council and the Constituencies have strong loyalties to these tasks and have appropxiate
professional autonomy in those roles, While also being accountable in amattix tbshion
to ICANN's staff management.
GNSO Counci1)s working methods
4.16 Looking at other aspects of the Council's Work. it is apparent that it is painstaking
and requires a lot of commitment from councilors. Data kindly collated and provided by
the GNSO Secretaxiat suggest that Counciloxs spend on average 12 hours per Week on
GNSO Work. Council interaction has tended to rely on regular teleconference calls and
intensive use of email (both public and private). There is relatively intense monthly
teleconference call schedule, With calls taking place at times that are necessaxily
inconvenient for some members scattered across the globe's time zones. Our analysis of
attendance at these calls reveals a remarkably high level of commitment. The
75

LSE Public Policy Group                                                                GNSO Review
teleconference call on Which our research team sat in Was very ski.llfu.l.ly and neutrally
modemted by the Chair of the Council. However, in interviews many panicipants
suggested that teleconferencing allows only limited opponunity for in-depth discussion
between Constituency membexs. Calls tend to favor the more outspoken panicipams
Who have English as a Iitst language.
4.17 The GNSO Council has made efforts in the last six months or so to increase the
frequency of face-to-face meetings. Council membexs told us that the format of the
fhce-to-thce meeting provides a thr more constructive opportunity for Constituency
membexs to sit and discuss policy Work (alien informally). It is alien the case that
informal `whiteboard sessions' can help to break down seemingly intmctable positions.
Council members referred to a common syndrome that Constituencies Will tend to
formulate positions early on in the policy process, and then stick to these positions in
the absence of an opportunity to be challenged on them or discuss options for
consensus. Our extensive analysis of Council public mailing lists has uneanhed many
examples suggestive of how emajl comespondence can alien perpetuate raLherin.flexib1e
positions and mildly confrontational exchanges. Face-to-tbce sessions can help to break
down Lhis syndrome.
4.18 Since Januzly 2002 there have been around 70 GNSO Council calls averaging just
less than two hours each. A high proponion of current and former Council members
suggested to us that it is not necessary to convene Council calls so frequently. They feel
that much ofthe Work and discussion that takes place on these calls is largely
procedural and could satisfactoxjly be canied out using Widely available online sodware
tools. In their view an unnecessary and somewhat Iituzlistic emphasis on frequent
teleconferencing With the Whole Council is not only extremely taxing on panicipams
(and off-putting for potential panicipants), but it also does not lend itself to soft
discursive communication required for effective issue analysis.
Recommendaiion 15
The GNSO Council should rely more onface-to-face meeting: supplemented by online
collabomifve method: ofworking The Chair should seek to reduce the use ofwhole-
Council Ieleconjhencing
76

LSE Public Policy Group                                                                GNSO Review
4.19 The human interaction between Councilors is an important pan of a successful
project management approach. We recommend therefore that the balance in
communications is shitied towards four or five face-to-face meetings per year, With
meetings over and above those held at ICANN conferences in accessible hub locations
(possibly West or east coast USA or Europe). This approach could be combined With an
intensive push to grow the use of online collaborative sottwaie tools. ICANN staff and
the GNSO have recently introduced new communication management tools, such as
Shinkuro, and pilots have received mixed but generahy positive responses. Listservs
could also be used for general communication between GNSO stakeholders. (At least
two collabomtive sottware designexs at the ICANN annual conference in Wellington
offered their development services free of charge for a year - see the transcxipt ofpubljc
forum for details.) We Would expect to see the GNSO1eading the line for ideas on how
to innovate With freely available collabomtive sottware tools.
4.20 It is an essential pxinciple of representation that the selection of representatives
should not be ljmited only to those Who can afford to bear onerous costs. The current set
of armngements Where GNSO councilors are not suppoxted to attend face-to-tbce
meetings reflects the paucity of ICANN's resources in its early days. It also causes
potential conflicts of interest problems, since it means that Councilors must search for
the Wherewithal to fljlflll their duties. (The only current exceptions are the councilors
selected by the Nominating Committee, Who are reimbursed.) ICANN's budget is due
to increase considerably in the next year and it should be feasible to meet reasonable
(that is, non-luxury) costs for travel and hotel accommodation only. Council members
Who are funded anyway might be allowed to donate their expenses allowance to their
Constituency's thuds or to other ICANN pro bono activities. Alternatively it might be
possible to set up a ring-fenced discretionaxy fund for the Chair of the Council to
allocate. We note also that the ICANN proposed budget for 2006-07 contains tinancial
provision to investigate and implement atmvel assistance policy for Supporting
Organization volunteexs.
Recommendaiion 16
The GNSO Cozmcilor: should have access to afzmdfor reasonable Imvel and
accommodaiion expense: xo attend designaied Council meelingg instead oflmving to
meel such co:I.vj?om their own resource: as ax present.
77

LSE Public Policy Group                                                                GNSO Review
4.21 The onerous costs imposed on Council members, particularly those from the NCU
Constituency, in attending ICANN armuzl meetings (otter; in for flung comers of the
globe) is an obvious barrier to participation. We heard arange of views about how to
tackle this issue. Some interviewees suggested that all Council members should receive
basic tmvel and accommodation expenses for a ljrnited number of meetings per year.
This may involve a signiticartt increase in expenditure for ICANN, and such
arrangements may need to be integmted over a longer term as part of a process of quid
pro quo. As an interim measure, We recommend that ICANN make available to the
GNSO an annual fund for tmvel and accommodation. and that Constituencies are able
to apply to the fund to cover a designated minimum attendance at face-to-tbce meetings.
This fund might be administered at the discretion of the Chair, and payments should be
agreed by vote in the Council. GNSO and ICANN may Want to consider moving in the
longer term to an armngement for covering all tmvel and accommodation expenses for
Councilors.
4.22 We tum next to the issue of Task Forces and GNSO's use of extemal expertise in
the policy development process. There is currently a juncture in the PDP, Where the
decision is taken to either launch a Task Force or to continue With ICANN staff
managers coordinating policy development by the Council as a Whole (see Annex A,
Figure A46). The distinction between these two routes is not clear to us. Some
interviewees suggested that the Task Force route is more suited to narrower and more
focused study. However, there seems to be some variation from this model. For
example, the most recent Task Force launched by the GNSO Council is a Wide-ranging
review of the purpose of the WHOIS database and privacy issues relating to data stored
on it.
4.23 We analyzed the membership of Task Forces (and indeed steering groups
overseeing Task Force Work) and found that the current mix of participation has largely
been ljmited to Council members and representatives from Constituency organizations,
see Figure 26 below (and Annex A, Figure A4S and ASO for more details). There is no
representation from external policy or technical experts and there has been very limited
representation from the private sector or governments. This approach has meant that
Task Forces have commonly demonstrated the same intmctable policy divides that have
been visible in the Council. Because many of the people on Task Forces are GNSO
73

LSE Public Policy Group                                                                GNSO Review
Councilors, they generahy help develop the Constituency policies themselves, and then
have to vote for the policies they have created.
Figure 26: The backgrounds of participants on WHOIS Task Forces since
February 2001
Council      Constituency   Other ICANN       Total
members        membexs            bodies
WHOIS Review 2001          5          10            4           19
WHOIS TF 1, 2, and 3               14                17                   3                34
Ma  2003
WHOIS Pxivacy Steeling       10                        2           IS
Grou| Ma  2003
WHOIS TF12 Jut zoo4             16 |          2                27
WHOIS TF123 Jan 2005       10          12           2          24
Total |artici|ant seats          55          54           13          122
Filled b  Total individuals           20                   30                      7                    57
NOTE: Other ICANN bodies consist of ccNSO representatives a1terZ003, General Assembly
representatives, and ALAC representatives.
same LSE ppa and as Dfattso Task Fame |anm|ams
4.24 By compaxison With other global and intemational bodies With policy development
roles, the reliance on GNSO-dominated Task Forces in the PDP seems to be a
somewhat i.nWard-looking approach. Annex C gives information on a number of
compamtor bodies to suppoxt this view and We have discussed vaxiations quite Widely
With interviewees. This inwardness is paxticularly marked because much of this policy
Work is inevitably linked to national and regional policies that are already in place in
different countries and zones across the World (like the European Union bloc). Panly
because of a tendency for Amexican and Angle-Saxon participants to be the most
influential members, some interviewees told us that GNSO outputs can sometimes seem
disconnected from the Wider World and from debates going on in more mainstream
international channels. Letters to the ICANN Chairman on the recent WHOIS
formulation from mqjor national govemment bodies (such as the UK Office of Fair
Trading) are cases in point.
4.25 Another example perhaps illustmting how GNSO policy output can follow its own
intemally dtiven logic is the tinal WHOIS Task Force report on resolving conflicts for
registxies and registraxs between compliance With national pxivacy laws and ICANN
79

LSE Public Policy Group                                                                GNSO Review
contractual obligations. This policy stipulates that national competition authoxities Will
have 45 days to provide their ruling in response to a condjct. More than one of our
interviewees suggested that it Was somewhat optimistic to expect any competition
authoxity to provide amling (as opposed to guidance), and even more optimistic to
stipulate that this be produced Within 45 days. The geneml opinion from other
interviewees is that ifpush came to shove, ICANN Would have to respect the laws of
the countty.
Recommendaiion 17
The GNSO Council should make more use ofTa.vk Force:. Task Force panicwarn:
should be more dfverse and should be dmwnjhzm a wider range of people in the
Jmemex community, and nwiomzl and imernaiionalpolicy-making communities.
4.26 Task Forces provide a structured Way ofpxizing open policy issues and developing
informed and focused issue analysis for review by the Council. They also provide a
usefijl oppoxtunity to integmte new faces and different sources of expertise into the
policy making process. Because the cument make-up of Task Forces has genemlly
mirrored the balance of constituent interests in the Council, their debates and outputs
too alien reproduce the same problems of intractable interests that are seen at Council
level. A Workable rule of thumb should be that Council membexs and Constituency
representatives should make up no more than half of future task forces. If six
Constituencies are retained as now, We do not believe that it is necessary for Task
Forces to include two representatives from each Constituency. A Task Force of 12
membexs could instead include no more than from each Constituency, plus (Where
appropxiate) representatives from ICANN Advisoxy Committees such as ALAC and the
GAC. Depending on the nature of the issue under consideration, the remaining 4 to 6
membexs could be chosen from other organizations from the pxivate and
intergovemmental sectors, ideally drawing in people With distinctive expertise and
information/analysis capabilities.
Recommendaiion 18
An ]CA]VNA::ociuIe stake}/[older caiegory ofpurlicwaiion should be Crewed :0 as to
create a pool of readily available external expertise, which can be drawn upon to
populaie Task Force: where relevant.
SO

LSE Public Policy Group                                                                GNSO Review
4.27 There is strong research evidence from many voluntaly organizations and
networks in many different countries and sectors that people get involved pximaxjly in
their operations because others Whom they know personally tell them about the
organization or ask them to join or support its Work. An ICANN consultant or associate
stakeholder category could also be established With a view to encouraging ongoing
paxticipation in the ICANN process by people With appropxiate technical or policy
expertise. Other international and global bodies Working in similar areas to GNSO make
extensive use of expert-dominated task forces and Jind that this approach allows for
much more expeditious and constructive Working on detailed policy development.
Many intexviewees stressed to us that it is important to recognize that task forces can be
organized in many different Work patterns, as opposed to the long-drawn out and multi-
stage process emphasized in GNSO Council at present. For instance, a pattem ofpaying
predominantly expext and committed people to come and conduct business intensively,
tbce to face and then to produce a comprehensive dmh policy for GNSO Council to
consider, might allow GNSO to become more expeditious and consensus] at the same
time. It could also be an effective Way of disseminating better information about
GNSO's processes and modes of operation to awide audience of contacts Who might
then become potential participants in GNSO Work more genemlly.
4.28 We tum next to an important aspect of the policy development process, the stage
When GNSO requests Constituencies to seek public comments on options and
proposals. Public comments valy considerably in their numbexs and volume (see Annex
A. Figure A51). Historically, the early public comments periods on the UDRP proved
compamtively active, With intellectual propexty interests debating With pxivacy and civil
liberty advocates. Much of this debate has continued through the public comments on
the recent WHOIS Work. By contmst, policy development Work on registry sexvices has
attracted very little public interest. The topic ofbxinging in new genexic top-level
domains has shown a slight increase in public interest. The available documentation
shows that GNSO's public comment pexiods have frequently involved individual
membexs of Constituencies resubmitting their views under the guise of `public'
stakeholders. Analysis of comments shows clearly that a high proportion of `public'
conttibutors are Constituency membexs. There may be undexstandable reasons for these
patterns, but it does tend to give the public comment pexiod a somewhat phony identity
ifljttle Widening out of debate occurs duting them.
S 1

LSE Public Policy Group                                                                GNSO Review
Voting in the GNSO Council
Figure 27: Proportion of all votes taken by members of the GNSO Council
showing disagreement, abstention, or `did not vote'
Agxeeing on PDP process (includes Task Forces or
committee as a whole)
Substantive policy issue or vote on policy
reeoxmmdatiors
Agxeeing on other process (e.g, appomtments, fimds,
thanks)
Ageeixg to pmgxess to the next stage ofa PDP
Request to the Board orto comxnnicate withthe Board
Foxma committee or group to consider anissue
0     50    100   150   200  250  300   350  400
Number of individual Councillor votes (Jan 2003 to October 2004)
D Abstain or abstain by proxy                  D Did not vote                  I No
same Lse ppa analym Dfamso damn amso cnunnl vntmg
4.29 Tuming to the Way in Which the GNSO Council itself opemtes, anumber of
inierviewees suggested that there is too much of a `legislative' approach ralher than a
deljbemtive approach, despite the best efforts of the Chair to try to move members
towards `consensus policies'. Cxitics argue that there are too many votes in the Council.
Our analysis shows that there are a large number of votes on proceduml and
administrative matters. For example, between Januzly 2003 and October 2004, 63 votes
Were held in 27 teleconference meetings, an average of more than two per convexsaiion
(see Figure 27 above). Many procedural/administrative votes are non-controvexsial and
therefore generate high levels of consensus amongst members. However, alimited set
of the large substantive policy votes generated intractable blocs of interests. The current
voting rule requires 66 per cent agreement for consensus policies, but its impact needs
32

LSE Public Policy Group                                                           GNSO Review
to be assessed against Weighted voting for Registries and Registrars, plus the strong
overlap of interests between three commercial user Constituencies interests (the
Business, IP and ISP Constituencies). Both our interviews and review ofvoting patterns
across large policy issues such as WHOIS suggested that these blocs can otten
undermine consensus-making. However, some interviewees also suggested that
alignments between Constituencies can otien shift depending on the issue. It Would
certainly be an oversirnpljtication to suggest that registration and user constituencies are
always ahgned against each other. Taken together these features appear to provide little
systematic incentive for opposing blocs to Want to tind common ground for consensus.
Mqjor blocs (ofRegistries/Registrars on the one hand and the cross-Constituency
grouping on the other) seem to view the super-mqority vote of 66 per cent as attainable
if they can succeed in attmcting arelatively few `swing' voters either from the Non-
commercial Users Constituency or some or all of the Nominating Committee members.
So the currently rather low `consensus' threshold may actually foster mqj or blocs or
may at least provide little incentive for Constituencies to Work from an early stage to
reach agreement.
4.30 When the GNSO in its current form Was established as a sepamte body for generic
names in 2002, some key irmovations Were made to the voting system Within the
Council. In particular, the gTLD registries and the registrars are the only stakeholders
Who are contractually tied to ICANN. Because they Were seen as vulnemble to having
their business models changed as aresult of GNSO deliberations and policy proposals,
and because registration stakeholders account for at least 90 per cent of ICANN
revenue, they Were accorded extm protection in the GNSO's institutional set-up.
Therefore the votes of the Registry and Registrar Constituency representatives count as
double-Weighted in formal votes. Thus although GNSO's normal operations are
conducted by counting voices amongst the 21 members, on formal ba].Iots there are in
fhct 27 votes cast (With double-Weights for the three registry and three registrar
members). When the original DNSO Was established steps Were taken to define the
level of support needed for it to make `consensus' policy, Which Was set as policy
commanding the support of two thirds or more of the Council. The GNSO adopted the
same policy consensus threshold. Thus the Registry and Registrar votes combined add
up to less than half (12) of all Weighted votes (27) and they are Well short of being able
to define `consensus' policies in the GNSO Council (IS votes).
33

LSE Public Policy Group                                                                GNSO Review
4.31 `Consensus' decisions reached may not be very meaningful in cases Where
opposing blocs cannot agree on an acceptable compromise. A recent 2006 example of
the vote on the purpose of WHOIS shows how a super-mg ority can take place, giving
the impression of consensus. However, as many people have suggested, the vote in
early 2006 alter the ICANN armuzl conference in Welljngton generated What is
effectively an unsustainable policy position. in that national governments and law
enforcement agencies Will insist on access to WHOIS database information, as Will IP
interests investigating potential abusive behavior relating to registered trademark
domain names. This interpretation also seems to be underpirmed by subsequent letters
from mqjor government and law enforcement agencies to the ICANN Chair.
4.32 The combination of the current Constituency system, the voting system, and the
PDP's somewhat formulaic and inflexible mechanisms for developing globally relevant
policy does seem to mean that GNSO policy-making is unusuz]ly vote-focused and
`legislative' in character, When set against some compamtor bodies. In our discussions
With officials from other global policy development bodies such as those listed in
Annex C, We found that voting Was actually extremely mre and that consensus Was
generahy the product of discursive and itemtive Work around pre-prepared policy dratts.
One senior ofticial explained: `We almost never take avote. . .We have processes in
place for avote to be taken. . .but in my live years of experience and the 20 or 30 policy
papers produced, We have hardly ever voted.' The trick is to develop policy positions
comprehensively and inclusively enough so that votes do not need to be taken. Project
management, issue analysis, inclusion of relevant experts, and giving adequate
opportunity to discuss issues in-depth face-to-face are all currently somewhat missing
elements of policy development mix covered in this Part. Where they can be properly
articulated, the experience of other bodies is that votes are `rubber stamping' mther than
`conthct resolving' devices.
4.33 Considemtion of the voting systems used in the GNSO Council forms the last
theme ofthis Part. But it is important here also to bring back into focus that the Terms
of Reference for this Review asked that the LSE consider Whether the current
Constituency structure should be amended in some Way to better reflect the diversity of
global opinion on generic TLDs. We have discussed the opemtions of the current
34

LSE Public Policy Group                                                           GNSO Review
Constituency system in Pan 2 above, but it is impoxtant to look also at the implications
of any recommendations on the number and foxm of GNSO constituencies for the Way
in Which the Council itself operates. In addition to interviewees' comments to us about
paxticular Constituencies, many of those cxitical of GNSO's current opemtions argued
strongly for different amendments to the Constituency system, mnging from minor
tinkexing to radical deconstmction. Given the detailed evidence reviewed in Part 2, We
formed the view that although the current six Constituencies Were created only seven
years ago, they already constitute something of a `legacy' structure Whose
contempomry mtionale is no longer clear. The cument structure has been made
inappropxiate by a mnge of subsequent changes - such as the resolution of the most
pressing intellectual propexty issues via UDRP, the changing chamcter of registration
businesses, developments in the secondary market for domain names, the mpid growth
ofweb hosting companies and other forms of e-commerce markets. The logic behind
creating the current six Constituencies has thus proved highly time-specific. It is not
clear to many potential stakeholders Where their interests tit Within this pattern, and the
structure of six Constituencies With double-voting for two of them creates armngements
that are relatively complex and off-putting. At the same time, having to have three
representatives for each of six constituencies makes the GNSO Council a relatively
large body, perhaps larger than needed to be able to handle effective policy
development and certainly contxibuting to the `legislative' style of operating noted
above.
4.34 Focusing on the sepamte Constituencies representing Business, Intellectual
Property and Intemet Sexvice Providexs (ISPs), awide mnge ofintexviewees argued to
us that the reasons for there to be three distinct constituencies are not now apparent. In
recent GNSO Council processes representatives from these three constituencies have
tended to adopt closely ahgned positions, and the three also hold an influential `cross-
constituency' joint meettng at ICANN conferences. Some of the BC membexs tend to
have relatively specific and open interests in IP issues (for example, Walt Disney, Time
Wamer, and TAGI). By contrast, many of our interviewees argued strongly that IP
interests are no longer a distinctive subset of business and commercial interests and
therefore no longer Waxrant a sepamte organizational division of their own.
35

LSE Public Policy Group                                                           GNSO Review
Recommendation 19
The current GNSO Constituency structure should be radically simplmed so as to be
more capable of responding to rapid change: in the Jmernel. The Constituency
structure should be clean comprehensive (covering allpotemiul smkeholders) and
flexible, allowing the GNSO to respond easily to the rapid change: in the make-up of
Jmemex smkeholders. We suggest a sel oflhree larger Constituencies to represent
respectively Registration interests, Businesses and Civil Society.
4.35 In tandem With Recommendation 5 above that individuals' and organizations'
pximary membership should be of ICANN as awhole rather than of specitic
Constituencies, We suggest creating a simpler Constituency structure With only three
main divisions
  Registration, including the cument Registties and Registmrs
Constituencies;
  Business Users, including the current Business, Intellectual Property and
ISP Constituencies; and
   Cfvil Society, including the cument Non-Commercial Usexs
Constituency, but also ordinaly domain holdexs, and possibly
individuals currently represented via the At Large Advisory Committee
of ICANN.
Once organizations and individuals have joined ICANN they Would choose to join one
of the three genexic domain names Constituencies above that most related to their
interests. The definition of the Constituencies should be inclusive, so as to ensure that
there are no gaps Where organizations and individuals cannot find an appropxiate
Constituency. We believe that the suggested reorganization here responds to multiple
pieces of evidence about how interests are cutrently organizing themselves Within
GNSO. Yet it also creates a structure that is simpler, balanced, clearer to explain to
potential membexs and time-proofed against fhture changes in the Intemet that are
cextain to occur. In place of ahighly time-speciiic structure that Part 2 demonstrates has
had great difticulties in adapting to changes over as little as seven years, the suggested
new structure can flexibly accommodate changes in the balance and Weights of different
sectoxs and types of involvement With Intemet policy issues.
36

LSE Public Policy Group                                                                GNSO Review
Recommendation 20
A reorganization ofGNSO Constituencies would also allow the Council to be made
somewhat smaller (we suggest 16 member.? and hence easier to manage.
4.36 Clearly the Council must represent diverse points of view, but a body of 21 people
is a hard one to convene in one place and to get agreement from. With a simplified
Constituency structure as above, the number of representatives per Constituency could
be axranged Within Wide range ofpermutations that should be for the ICANN Board in
consultation With GNSO to decide. However, as anindicative guide We Would suggest
that allocating 5 seats to Registration, 5 to Business Users and 3 to Civil Society, along
With 3 Nominating Committee seats Would reduce the overall membership to 16 but yet
preserve the current balance of interests on GNSO Council. (It Would be possible to
allocate more seats to Civil Society, but there are already indications that perhaps it
Would not be easy to 6.1.1 them.) For ordinaly votes, not using Weighted voting, the Chair
Would gain a casting vote in the event of a tie, reflecting the recommendation above to
increase his/her role.
4.37 Moving on to the Council's voting rules, the two thirds mqjoxity for achieving
formal consensus policies seems somewhat too low, given the cument mix of interests
in the Council. At present Weighted voting creates some avoidable complications in the
GNSO Council. For instance, alot ofnormal business and the early stages ofaPDP is
done by `collecting voices'. But a mqjoxity painstakingly assembled in this Way by the
Chair can then be easily ovextumed later on in a Weighted vote, Where a different
calculus of suppoxt applies. The result is alot ofurmecessaly maneuveiing, plus
resentment among those Councilors not erg oying Weighted votes. Clearly there is aneed
to provide assurance to registties and registmrs that their interests Will not be ljghtly or
inadvisably affected by new PDP decisions. But We do not believe that Weighted voting
is the best Way of doing that. As one experienced ICANN paticipant obsexved on the
cument armngements in the Council: `[Weighted voting] Was, and still is, a symptom of
much deeper set of problems. . . and certainly not the solution'.
Recommendation 2]
The defnilion of achieving a consensus should be mixed to 75 per oem. Weighted
voting should be abolished. Both measure: could he? to create more incemfvesfor
dwrerom Constituencies to engage comlmclively with each when rather than simply
37

LSE Public Policy Group                                                                GNSO Review
roilemiing a (bloc 7 position in hope: ofpicking up enough uncommitted vexes :0 as to
win.
4.38 The voting system should be set up in such a Way that achieving broad consensus
is in the interests of all stakeholdexs from the start of policy development, regardless of
the issue. With a Council of 16 membexs, and Without any Weighted voting for the
Registration Constituency, We suggest that the consensus level be set at 12 votes (75 per
cent). Note that Weighted voting Would not be needed here to protect the undoubtedly
salient interests of registxies and registraxs, because the Registmtion Constituency
Would already have ablocking vote of 5 (44 per cent). Accordingly, any new consensus
policy Would require the assent of at least one Registmtion Constituency representative
to be approved. Note also that With a Council of 16 and 5 Business User votes this
Constituency too could also block. (If for any reason Weighted voting is still retained,
consensus policy-making Would be best fostered by allowing a block by Business
Usexs. With a GNSO Council of 16 and aRegistration Constituency of 5, retaining
Weighted voting Would give a total of 21 votes, ofwhich 10 Would be in the
Registration Constituency. In this case, We recommend that the level of suppoxt needed
for consensus policy should be mised to 17 votes [S2 per cent], so that With 5 votes
Business Usexs too Would have a capability to block.)
4.39 The GNSO Council elects two Directoxs (Seats 13 and 14) to the ICANN Board at
least evexy three years, and councilors also elect the Chair of the Council. In both cases
each member of the GNSO Council casts one vote in the Jixst round for a preferred
candidate (With no Weighted voting). If there is a mqj oxity Winner, then there is no
second round. If no mqj oxity Winner is produced in the Iitst round, the top two
candidates are sepamted oft, and a second round takes place. The Winner is decided by
simple mqjoxity. Under this system Counciloxs Whose candidate progresses to a second
round may vote for the same candidate in both rounds. Counciloxs Whose candidate
does not progress to the second round must vote for another candidate (or abstain). This
system does not create strong incentives for candidztes to `reach out' to people from
other Constituencies.
Recommendation 22
The way in which the GNSO Council vexes to elect Ma Director: to the JCAMVBOWJ
should be Changedxo we the Supplementary Vote system.
SS

LSE Public Policy Group                                                                GNSO Review
4.40 We therefore recommend introduction of a simple multi-preference system called
the Supplemental;] Vote cumently used to elect the executive Mayor of London (also
called the `instant run-oft` system in the USA). Here each Councilor has aballot paper
Where they can express two preferences, one for their first preference candidate and
another for a second preference candidate. In electing the GNSO Chair (a single oftice
holder) first preference votes are counted and any candidate With mqjoxity suppoxt is
elected. If no candidate has mqjoxity suppoxt, then the top two candidates (let's call
them A and B) stay in the race but all others are eliminated. We look at the second
preferences ofvoters for the eliminated candidztes, to see if any are for A or B, and if
so they are added to that candidate's pile. The Winner as Chair is Whichever of A and B
now has most votes. In electing two Board membexs, if two candidates have mqjoxity
suppoxt in first preferences they are elected and the process ends. If no candidate has
mqjoxity suppoxt in the first preference votes, then in the Supplementary Vote system to
select two Wixmers then the top three candidates (A. B and C) stay in the race and all
others are eliminated. We next look at the second preferences of voters for eliminated
candidates, adding any for candidates still in the race to their respective piles. The
Whmexs are now the two most popular of A. B and C on the basis of their combined first
and second round votes. The remaining Board seat is then Whichever ofB and C now
has most tixst and second preference votes. A slightly more complex case for the Board
member election is When one candidate A has mqjoxity suppoxt on first preferences:
here A is elected and their votes are removed from the election. The top two remaining
candidates (B and C) stay in the contest and all other candidates are eliminated. We
look at the second preferences ofthe eliminated candidztes and if any are for B or C We
add them to their pile. The last Board seat goes to Whichever ofB and C now has the
most Iitst and second preference votes.
Achieving a coherent vision for policy development
4.41 There are very few (if any) organizations in the World Which combine such a range
of different functions in the Way that ICANN does. By implication the GNSO shares
this somewhat sui generi: profile. It is a body charged With developing policy that
incorporates the views of global society as a Whole. It also contains and represents the
stakeholders that account for over 90 per cent of ICANN revenue. It must somehow
39

LSE Public Policy Group                                                                GNSO Review
develop pmcticable policies based on consensus of a Wide yet at the same time rather
namaw range of interests and bodies. Our interviews have shown a general consensus
that there is greater benetit to be gained from having all relevant stakeholdexs involved
in the policy development process, than not. A number ofintexviewees have suggested
that the registration Constituencies (both Registties and Registmrs) should be separated
off from the GNSO, on the grounds that their business interests generate obvious
conflicts of interest in their participating in policy development. A strong countexview
has been that ICANN is not a regulatory body (although it has regulatory aspects) and
so it is not the role of the GNSO to develop regulatoxy policy.
4.42 The changes suggested here address the problem of how to interpret and control
for this mnge of interests, some ofwhich claim to represent Wide sets of stakeholders
and some of Which are clearly based on narrower business interests. There are alien
predictable positions taken by Constituencies, and equally predictable differences
between them, and therefore fashioning consensus (or something resembling consensus)
is gigantic challenge. As one Board member suggested: `it leaves me With uncertainty
about how to achieve bottom-up consensus policies When the paxties Who help to
formulate these policies have fundamental conflicts'. This chimes With many other
views that We heard from intexviews, basically pointing out that most Constituencies
have too much at stake both in terms of day-to-day business and in texms of personal
ljvelihood and status to concede on issues Where they Would be better off either
pexpetuzting a status quo position, or sticking dogmatically to an intractable stance.
4.43 This syndrome presents problems for the ICANN Board in terms of its ability to
interpret Council recommendations and make a judgment on Whether to adopt them. On
policy development such as the new registty services PDP, Which involved some
procedural ground rules for reviewing changes implemented by gTLD Registties, issues
may be relatively uncontrovexsial and therefore relatively easy for the Board to agree.
Any policy issues, Which are anything more than uncontroversial or completely Watered
down Will inevitably bear the predictable signs ofpaxtisanshjp. As another Board
member told us: `the Board Will receive divexse interests that it Will now have to resolve
or reconcile. . .it is un.lj.ke1y that the Board Will get from its vaxious policy initiatives a
coherent view that doesn't contain some of the inescapable conflicts that these groups
exhibit'.
90

LSE Public Policy Group                                                           GNSO Review
4.44 To effectively countemct these difliculties We believe that applying a coherent
vision (]j.ke that behind the Whole package of measures set out here) is needed. The
GNSO Council should have a simpler, more inclusive and more undexstandable set of
Constituencies, one that can flexibly accommodate changes in the ICANN community
and environment. The Council's internal opemtions should not be clouded by the
complications ofweighted voting. Instead all parties should know from the start that a
demanding consensus level (75 per cent) must be reached for a decision to be made.
The Registration and new (integrated) Business constituencies should both have a
capacity to block change that does not elicit their agreement. The involvement of all
other usexs should be simplitied and the base of interests represented here should be
broadened. The GNSO Council should become a somewhat smaller body and its
delibemtions should take place overwhelmingly fbce-to-face. It should focus its
activities on an expanded number of meetings per year, With less use of
teleconferencing. The Council should also focus on making key policy decisions and it
should delegate detailed Work more, making stronger use of more diverse and expert
task forces to bong in expextise and to help speed up policy processes, as Well as using
the ICANN staff suppoxting its deliberations more constructively at all stages of the
process. Our illustrative examples above show how these guiding pxinciples might be
achieved, and how the different elements can Work together harmoniously. But there are
many other possible combinations that could also be made to Work consistently
together.
9 1

LSE Public Policy Group                                                           GNSO Review
Part 5:
The regularity of the GNSO's operations in
complying with ICANN's Bylaws and operating
procedures
5.1 In this short Part We review evidence on the extent to Which the GNSO Council
and its Constituencies are encouraging and maintaining an environment of policy
compliance. In paxticular We look at the extent to Which:
-    GNSO complies With ICANN Bylaws and operating procedures;
-    GNSO complies With its own rules and procedures;
-   There are sound ljnkages between the GNSO policy development
process and the ICANN policy compliance program;
-   ICANN and the GNSO have Worked to follow up on the impact of
previous policy development Work.
We have not looked here at issues concerning the compliance ofregistmrs and
registties With ICANN operating contmcts, Which is outside our research scope.
5.2 The evidence We have collected and our many intemctions With GNSO Council
membexs and ICANN staff all show that they are clearly concerned to operate
Within the ICANN Bylaws and operating procedures. The ICANN Bylaws set out in
detailed and prescxiptive language the stmctuml and procedural parameters of the
GNSO, including the Policy Development Process. We asked our survey
respondents, both Constituency membexs and individuals, to give us their views on
the extent to Which the GNSO compljes With ICANN Bylaws and operating
procedures. Figure 23 below shows arelatively high degree of convergence on the
view that the GNSO is compljant in it Work (taking into account the relatively small
number of organizations responding for some Constituencies). However, gTLD
Registties and a group ofindividuzls show signs of disagreeing With the views that
the GNSO is compljant.
92

LSE Public Policy Group                                                           GNSO Review
Figure 28: Perceptions of survey respondens on the extent to the GNSO
complies Mth ICANN bylaws and operating procedures
g so
WE zs
i zo
mg  Is
Jg  to
g   5
Z   o
gTLD       Regtsnrs        Be          NCUC          ]:p           Isp           IND
Regtstnes
D High (scores 5 to 7)      D Mid (scores 4)      I Low (scores 1 to 3)      D Not sure
NOT]; IND I mdmduals
5.3 We examined What factors might cause four out often gTLD Registties
responding to our survey to take the view that the GNSO shows low levels of
compliance With ICANN Bylaws and procedures. Elsewhere in our suxvey We asked
respondents to give us a score on aLi.kert scale from 1 to 7 for different aspects of
the Policy Development Process (see Annex A Figure A54 for alist of these aspects
and Annex E for full results). Figure 29 below gives a summary of the highest and
lowest ranking aspects for each Constituency. These results suggest that the gTLD
Registties have mqjor concems With the Way that the GNSO scopes its policy Work.
Interviewees from this Constituency also suggested strongly to us that the GNSO
(and indeed ICANN as awhole) has tended recently to venture out of scope and
interpret cextain issues, particularly those relating to Registry contmcts, as policy
relevant. The current PDP Work on amendments to conditions of existing contracts
With gTLD Registxies has led to a high level of controversy about Whether or not the
GNSO should be allowed to develop policy in this area.
5.4 It is perhaps not too sutpxising that gTLD Registties have concems over the
scope of the GNSO policy development Work, given the obvious business
implications of allowing the GNSO to amend their contmctuzl conditions. In our
interviews, expexienced ICANN paxticipants have suggested that scope issues are
otien used instmmenta]ly to deflect and delay issues. As mentioned above, the
93

LSE Public Policy Group                                                                GNSO Review
Figure 29: Constituency survey response on how well the GNSO performs on
different aspecs of its policy development work
RANKED                                      RANKED
GNSQ                                   HIGHEST                                  LOWEST
Cunsmwznc
gTLD                   Making the best use afpalicy             Scoping policy work
Re : ~ Sujes                    supp art resources                                              appropriately
Registraxs            Delivering practicable                      Making use of extemal expertise
recommendations to the ICANN        and research
Board
BC                      Picking the right is sues for                Sticking to agreed time
development                                   schedules
NCU                Identifying issues early enough        Making use of extemal expertise
and research
Ensuring that the PDP inc axparat/es     Sticking to agreed time
the widest practicable range of           schedules
views
ISP                                                           Inmmczent data avmluble
NOTE: In our online survey we asked Constituency members to score how effectively the GNSO
pexformed on different aspects ofits policy development work. The list of aspects is available in
Annex A, Figure A44. Scores were given on a Likert scale from 1 to 7, where 1 I Not at all
effectively and7 I Very effectively.
same Lse ppa mama sum|  Dfcnnsumem members
Council and ICANN staff engaged in vaxious discussions around the scope for the
new PDP on contmctuzl conditions. Also scope issues have framed the recent
discussion on the purpose ofthe WHOIS database, in this case relating more to the
scope ofthe database itself rather than to the coverage ofthe policy Work. In
general, registties and registmrs support much namawer interpretations of GNSO
scope. This translates into similar views about the scope of ICANN's activities as a
Whole. Some people argued that the GNSO has taken on too Wide a concept of its
responsibilities, seeking to pronounce on issues that affect the Whole of ICANN and
Without co-ordinating its Work sufficiently With other suppoxting organizations or
pans of ICANN (like the Govemment Advisoxy Committee). They also feel that
GNSO has declared too many PDPs at once, Without achieving resolution on
ongoing issues. However, as the table above and more detailed data in Annex E
suggest, most Constituency members do not see scope ofthe GNSO as too much of
problem: they believe that the scope of ICANN and the GNSO responsibilities are
essentiahy compatible. This view is continued by awide mnge ofintexviews across
other Constituencies. We recommend in Pan 3 that the GNSO should develop a
more formal expression of its upcoming policy Work (possibly covexing 2 years),
94

LSE Public Policy Group                                                                GNSO Review
and We believe that many current concerns about possible over-scoping could be
ironed out in the process of thinking about such a planning document.
5.5 Another area Which mises questions about the extent to Which the GNSO is
compljant With ICANN Bylaws concerns sticking to time schedules. As We discuss
at length in Part 4, a Wide range ofinterviewees argue that the GNSO Council
complies With the pxinciples of the ICANN Bylaws and the key steps on the PDP,
but does not comply With the currently unreahstic time-fmme set out in the Bylaws
for the PDP. We Would take the view that the PDP could benetit from a more
relaxed and more output-focused approach to time management, maintaining some
key delivexy mjlestones in order to ensure the structural integrity of the process.
Either Way We argue above that it is not helpfijl to continue With the current mix of
high regard for ICANN Bylaws, unreahstic timeframes in the Bylaws, and
significant time ovemm on policy development Work.
5.6 The constitutional relationship between the GNSO and ICANN is fmmed by the
ICANN Bylaws and the GNSO Rules of Procedures. The cument GNSO opemting
procedures run to just under 3,000 Words. As one Would expect, they contain
detailed provision covexing structures and process relating to the GNSO. The
cument passages in the ICANN Bylaws relating either specitically to the GNSO or
Supporting Organizations runs to about 6,000 Words. There is considerable
duplication between the two documents, paxticularly on issues relating to the PDP.
A small but significant minoxity ofinterviewees have questioned the need to have
the PDP procedures for the GNSO Wxitten out in full in both documents. In most
legal-constitutional and legislative contexts, it is generahy the case that the volume
ofprescxiptive detail gets heavier as one moves down institutional mngs or levels.
Most national legislation for example Would not contain such detailed prescription
that one Ends in the PDP section of the ICANN Bylaws. Instead, this kind of
prescription Would be left to statutory instruments, executive orders and so on. The
implications of this point for the GNSO is that by removing the detailed provision
on the PDP from the ICANN Bylaws it Would free up the GNSO Council to amend
or innovate With more project-based Working, Without the stigma of contmvenjng
top-level constitutional rules. The Bylaws themselves could refer to the PDP or to
95

LSE Public Policy Group                                                                GNSO Review
fostexing bottom-up policy development, but in terms of the core pxinciples
involved and the main objectives to be achieved.
Recommendation 23
The amount ofdeIajledpre:cr?Iive provision in the JCANN Bylaw: relaxing to the
operation: oflhe GNSO should be reduced JCANN Bylaw: should outline broad
princwle: and objectfvesfor the GNSO but the detailed operational provision
(including the section on the PDP) should be Immjierred to the GNSO Rule: of
Procedure. This would allow the GNSO to agree amendment: and to introduce new
irmovaxiortv in its working method: and Iimeline: in a more realistic andj'lexible
way, while operating within JCANN 7: guiding princ?le:.
5.7 We have canied out this Review at an interesting juncture in terms of the
relationship between the GNSO and the ICANN Board. The provisional settlement
With Vexisign Inc. (still to be completely Iinahzed at the time ofwxiting) has
undoubtedly caused controversy amongst the ICANN community at exactly the
time that our Iieldwork took place. Although this decision is not Within our study's
scope, it has led to Wider questions in our research on how a `consensus policy'
passed by the GNSO Council should be treated by the ICANN Board. This question
certainly has direct implications for perceptions of compliance across the ICANN
community, as Well as for people on the outside looking in. In our survey, We asked
Constituency members to give a score for the extent that ICANN (as distinct from
the GNSO) complies With its own bylaws and procedures. Figure 30 below gives a
summary of responses by Constituency. Although there is geneml agreement that
ICANN does follow its own rules, each Constituency shows apocket of skepticism
(low score from 1 to 3). In our interviews some people on the Council took the view
that if GNSO reaches a consensus decision then it should be binding on the Board
and cannot be ignored or overtumed. Others argued that the Board must not ljghtly
reject a policy developed by GNSO and meeting the two thirds ofweighted votes
cxitexion, but must explain and offer reasons. But equz].ly they argued that the Board
has to take account of the views of all the Supporting Organizations plus other
bodies (such as the Government Advisoxy Committee) in reaching an overall
96

LSE Public Policy Group                                                           GNSO Review
Figure 30: Perceptions of survey respundens on the extent to which ICANN
follows its own bylaws and procedures
 |
|
 |
|   I
 |             I
|
|   5 |             I
ITITI |    I
gTLD       Regtstrars        BC          NCUC          IP           ISP           IND
Regxstnes
a mg}. (scores s to 7)      m Mid (scores 4)     I Low (scores 1 to s)      a Not sure
ICANN view on complex issues spanning beyond genexic domain names alone in
their implications.
The impact of GNSO)s operations
5.8 Pushing this point in the previous paragraph slightly further, Figure 31 below
shows responses from Constituency membexs in our online suxvey on the extent to
Which ICANN Board takes into account the views of particular Constituencies.
5.9 Vexy little Work has been done by ICANN or the GNSO to follow up on the
effectiveness of consensus policies that are implemented. Vexy few of the early
consensus policies on procedures covexing transfer of domain names from one
registrar to another, or procedures for deleting domain names from registry files,
have ever been measured for impact or compliance. A question that axises is
Whether GNSO policy is specific and mainstream enough for any follow-up or
measurement Work to be done? Policy on the transfexs and deletions of domain
names might Well provide opportunities for implementation studies. However, much
of the WHOIS Work is vexy genemlized and peiipheml for sexious impact study
Work. For example, the policy covexing conflicts between national laws and ICANN
contractual obligations relates to decisions, Which, in the Words of one senior
ICANN official, `may never in actuahty have to be taken'.
97

LSE Public Policy Group                                                                GNSO Review
Figure 31: Perceptions of Constituency survey respondents on the extent to
which the ICANN Board take into account the views of their Constituency
Is
a  14
|  lz
|  to
kg   s
E   6
|   4
Z   z
o
gTLD Regtstnes     Regtstmrs            Be              NCUC               ]:p                Isp
D High (scores 5 to 7)       Q Mid (scores 4)      I Low (scores 1 to 3)       D Not sure
5.10 None the less a broad mnge of suxvey respondents and interviews raised the
issue of impact assessment Work. There has been no empixical analysis of early
policies relating to the accumcy of WHOIS data. The Geneml Accounting Office
(GAO) undextook an emphical study of the accumcy of WHOIS data, Ending that
around 6 to 7 per cent of data Was incorrect or missing. This study Was relatively
limited however in its sample, and there is clearly scope for Ihrther empixical Work
in this area. Encouraging the registration industty to evaluate levels of accuracy of
WHOIS data is not an easy prospect given the high costs and logistical cha]lenges
that are involved. Our review of the GNSO mailing lists over recent yeals reveals
somewhat tense and confrontational emajls between ICANN staff and GNSO
representatives from the registration Constituencies. ICANN has discussed the
possibility of introducing gmded penalties for registration stakeholders for
inaccurate WHOIS data records (in foot this Was suggested in the previous informal
GNSO Council review). It is not clear to us to What extent progress has been made
in this area.
5.11 Following up on the impact of the GNSO policy development Work should not
necessaxjly be limited to large empixical studies of foxmahzed policy. Policy Work
does not have to be judged exclusively on the number of pages of text produced or
the number of organizations complying With procedures (although this is one very
impoxtant aspect). Much of the value of the GNSO and its style ofpoljcy
development lies in highly regulaxized and inclusive discussion around issues that
are technical and multi-faceted. Participants in the policy development process are
93

LSE Public Policy Group                                                                GNSO Review
highly expexienced and Well-qua]jIied, and so the intensity and detail of the
discussion comes at a high premium. (We should not forget that the GNSO has
devoted over 30,000 hours in total to the subject of WHOIS). The result of this
should be that the GNSO (and ICANN) should be attempting to evaluate the impact
of its policy development Work in Wider circles, and over time Working on the
assumption that impact can be measured in lots of different Ways. For example,
many pxivate sector organizations now devote signiticant resources to measuring the
asset value oftheix Website, beyond simple financial projections. A similar approach
might be taken by the GNSO, focusing on `Mm is the value-added of our policy
development Work?'
5.12 We demonstrated earlier (in Part 2) that the GNSO has chIonica].ly low
visibility outside the ICANN community. We also asked Constituency members to
rate the impact of the GNSO and its policy Work on government and international
policy making organizations, and on commercial organizations. Figures 32 and 33
suggest that they feel that the GNSO has little or no impact. Our struggle to canvas
opinion about the GNSO across Wider commercial and non-commercial sectoxs has
tended to reinforce such low estimates of its influence. Nevertheless, the policy
development activity of the GNSO, in areas such as new gTLDs and IDNs, not to
mention ongoing hot issues around access to pexsonal data through the WHOIS
database, has direct (if small) relevance for avast range of individuals and
organizations using and providing sexvices relating to the Intemet. More formally
measuring the extent to Which the GNSO and its Constituencies are mfluencing
discussions and debates should therefore be a high pxioxity for ICANN.
99

LSE Public Policy Group                                                           GNSO Review
Figure 32: Perceptions of Constituency survey respondents on the extent to
which the GNSO influences governments and international policy makers on
rdevant issues
14
|  1Z
|  1o
|  s
|   6
|   4
Z   z
0
gTLD Regxstnes      Regxstmrs               BC                 NCUC                  IP                    ISP
D High (scores 5 to 7)      I] Nljd (scores 4)      I Low (scores 1 to 3)      D Not sure
Figure 33: Perceptions of Constituency survey respondents on the extent to
which the GNSO influences commercial organizations on relevant issues
14
|  1Z
|  1o
|  s
|   6
|   4
Z   z
0
gTLD Regxstnes      Regxstmrs               BC                 NCUC                  IP                    ISP
D High (scores 5 to 7)      D Nljd (scores 4)      I Low (scores 1 to 3)      D Not sure
Recommendaiion 24
Both JCANN and the GNSO Council shouldperiodimlly (my once everyffve year.9
compile or commission uformul (quamilaifve and qualilaiive) assessment ofxhe
influence oflhe GNSOX: work on developing policy for generic name:. This should
include an analysis ofhow the GNSOX: influence with national government.`;
imemwional bodies and the commercial sector might be extended
100